HyperBridge Platformhyperbridge.digital ↗
QuantumOS
Book a demo
Features

13 pillars. One platform. One ledger.

13 feature pillars and 105 capabilities, each described below as it works today, including where it stops. Plans are flat monthly fees with no commission on your sales, and every plan includes the full platform; see the pricing page for the details.

storefrontStorefront X3A fast, cached storefront that runs on Cloudflare Workers, with a visual page builder, multi-country settings and mobile apps.Deep dive →posPoint of SaleA browser-based register for your counter, with cash-up, payment-machine support and a queue that keeps sales safe when the connection drops.Deep dive →opsOrders, Inventory & WarehouseReserve stock at checkout, route orders to the right location, and run receiving, picking and transfers from one system.Deep dive →marketplaceMarketplaces & ChannelsRun your own multi-vendor marketplace, and send your catalog to Google, Meta, Amazon, Flipkart and other channels.Deep dive →loyaltyLoyalty, CRM & MarketingPoints, tiers and referrals, customer segments, and message journeys, all run on your own customer and order data.Deep dive →platformPlatform & Developer ExperienceA scoped REST API, OAuth, signed webhooks, an MCP server and a developer portal for building on your store data.Deep dive →sol-orbitSOL ORBIT Cache ArchitectureSun, Origin, and Luna: a cache control plane for edge delivery, tenant truth and freshness.Deep dive →aiAI & IntelligenceAI that works from your orders, catalog and customers, with your own model keys and rules you control.Deep dive →conversationalConversational CommerceChat with customers on WhatsApp and Telegram: browse the catalog, add to a cart, get a payment link and be handed to a person.Deep dive →bridgePartner Bridge & MigrationA signed, retried event feed to partner products, plus importers to move a catalog, customers and orders onto QuantumOS.Deep dive →eventosEventOSSell tickets, take registrations and check people in for events and cinema screenings, in the same system as your store.Deep dive →hotelosHotelOSRun rooms, rates and reservations for one or more properties in the same system as the rest of your business, and take bookings on your own website through a booking API.Deep dive →b2bB2B CommerceGive trade customers their own portal to request quotes, reorder, route purchases for approval and track invoices, while your team runs pricing, credit and orders from one back office.Deep dive →
Standalone product · Early preview

ShiftMotive

A standalone staff operations product powered by QuantumOS X3. Shopify connection is planned, not live.

Explore ShiftMotive →

Storefront X3

storefront

A fast, cached storefront that runs on Cloudflare Workers, with a visual page builder, multi-country settings and mobile apps.

Storefront X3 renders your shop on Cloudflare Workers and serves repeat visits from a cache that is cleared by product, category or page when something changes. Merchants build and publish pages in a visual page builder, show or hide blocks for chosen audiences, and run experiments. The same storefront can sell in several countries and currencies, show product text in nine Indian languages, install to a phone's home screen, and be packaged as a mobile app. The Value Store moved its catalog off OpenCart onto it, and several brands run it today.

Shared or dedicated deployment

Architecture

Run on a shared multi-business Worker with your own domain and theme, or on a dedicated Worker of your own.

Storefront X3 is a Next.js application built for Cloudflare Workers with OpenNext. Many businesses can share one deployment, each with its own domain, theme and data, and a business that needs its own deployment can have a dedicated Worker. Either way the storefront reads the same catalog, prices and stock as the rest of QuantumOS.

  • Cloudflare Workers runtime, built with OpenNext
  • Shared deployment with a domain and theme per business
  • Dedicated Worker available for a business that needs one
  • One catalog, price and stock source shared with the admin and POS

Edge caching and targeted purge

Caching

Pages are cached close to the shopper and can be purged by tag, so a change reaches visitors without clearing the whole cache.

Responses carry cache headers that let the edge serve a stored page while it is refreshed, and keep serving it if the origin errors. Pages are tagged, so a soft purge targets only the pages affected by a change. Requests for the same uncached page at the same moment are coalesced into one request to the origin. The admin has a manual purge for the moments you need it. In a scripted check on 5 July 2026, The Value Store's cached home, shop and category pages answered at a median of 67 to 77 ms and a 95th percentile of 112 to 151 ms from one test location. That is one business measured from one place, not a guarantee for every shop.

  • Cache headers with stale-while-revalidate and stale-if-error
  • Cache tags and soft purge by tag
  • Concurrent requests for an uncached page are coalesced into one origin request
  • Manual purge from the admin
  • Measured 5 Jul 2026 on thevaluestore.in: cached pages 67 to 77 ms median, 112 to 151 ms 95th percentile, one test location

Visual page builder

Marketers assemble pages from blocks in a visual editor. Each block's settings are checked before the page is published.

Pages are built from a library of blocks, and every block declares the settings it accepts, so the editor shows the right fields and bad values are caught before they reach the storefront. The editor keeps an undo and redo history while you work. Publishing saves a version of the page. Landing pages can be created as pages of their own.

  • Library of blocks with checked settings
  • Undo and redo while editing
  • A saved version each time a page is published
  • Campaign landing pages as separate pages

Audience rules on blocks

Show or hide a block for a segment, a customer attribute, a date window, a country or a device type.

Each block in the page builder can carry rules. An audience rule matches a customer segment or an attribute, and can combine conditions with and, or and not. A visibility rule limits a block to a date range, a geography or a device class. The page renderer evaluates the rules when it builds the page.

  • Audience rules by segment or customer attribute, with and, or and not
  • Visibility rules by date window, geography and device
  • Configured in the editor, with no code change

Countries, currencies and languages

International

Country settings carry the currency, default language, address format, delivery partners and payment methods for a market.

The storefront has country profiles for India, the UAE, Singapore, the UK, Malaysia and Bangladesh, each with its currency, default language, address format, delivery partners and payment methods. Shoppers can switch the display currency, and prices are converted with exchange rates from a public rate feed that are cached for six hours. If the feed is unavailable, a stored set of rates is used so conversion keeps working. The interface supports a long list of languages, including right-to-left scripts such as Arabic.

  • Country profiles for India, UAE, Singapore, UK, Malaysia and Bangladesh
  • Currency, language, address format, delivery partners and payment methods per country
  • Display currency switcher with exchange rates cached for six hours
  • Stored fallback rates if the rate feed is unavailable
  • Right-to-left languages supported

Product translations

New

Store product names and descriptions in nine Indian languages, with an AI draft button for each product.

Each product can carry a name and description in Hindi, Tamil, Telugu, Bengali, Kannada, Malayalam, Marathi, Gujarati and Urdu. The storefront shows the translation when a shopper picks that language and falls back to English when none exists. An AI button drafts a translation for one product at a time, keeping brand names and model numbers as they are. The draft is saved as an AI translation that you can edit afterwards. It is not held back for approval, so read it before you rely on it.

  • Nine Indian languages for product names and descriptions
  • Falls back to English when a translation is missing
  • AI draft for one product at a time
  • Drafts are saved as AI translations, are editable, and are not held for approval

AI-powered storefront search

New

Search that tolerates typing errors, expands your synonyms and falls back to meaning-based matches when keywords find nothing.

The storefront uses Kynetra Search, described in full on the AI page. Queries are cleaned up and matched with fuzzy text matching, your synonyms are applied, and results are ranked by weighted signals. When keyword search finds little or nothing, a meaning-based lookup can suggest the closest products. Search activity feeds Kynetra Prime, which proposes improvements for your approval.

  • Fuzzy matching and merchant-managed synonyms
  • Meaning-based fallback when keywords find nothing, with keyword results if it is unavailable
  • Redirects, pinned products and zero-result fallbacks set by merchandisers
  • Search events feed Kynetra Prime's suggestions

Installable web app

Mobile

Shoppers can install the storefront to their home screen from the browser, with no app store.

The storefront ships a web app manifest and a service worker, and an install page explains how to add it to the home screen on each device. When the browser offers an install prompt, the page can trigger it. Once installed, the storefront opens full screen. Catalog and cart requests always go to the network, so prices and stock are never shown from a stale offline copy.

  • Installs to the home screen from the browser
  • Install page with instructions for each device
  • Opens full screen once installed
  • Catalog and cart requests always use the network, so stock and prices are current

A/B testing

Compare two themes for real, or test page-builder variants and see the probability that a variant beats the baseline.

Theme tests split visitors between two themes. Each visitor is assigned deterministically, so the same visitor keeps the same theme, and results are compared with a significance test on add-to-cart, checkout and purchase goals. Page-builder tests record conversions with their revenue value and report the probability that a variant beats the baseline. You can pause, resume and end an experiment from the admin. One theme experiment runs at a time for a business.

  • Deterministic assignment, so a visitor sees the same variant on return
  • Theme tests judged on add-to-cart, checkout and purchase goals
  • Page-builder tests record conversion revenue and the probability of beating the baseline
  • Pause, resume and end from the admin
  • One running theme experiment per business

Mobile apps

Mobile

Generate an iOS and Android app project for your store, configured from the admin, that loads your live storefront.

A Capacitor-based app shell is generated per business from settings in the admin: name, icon, splash screen, bundle ids, deep-link scheme and push settings. The app loads your live storefront, so catalog and theme changes appear without a new app version. Push tokens register against the same customer record, and a version gate can require an update. The build produces native projects that your team opens in Xcode or Android Studio to sign and submit to the stores. Signing and store submission are not done for you.

  • Configured in the admin under Settings, Mobile App
  • Loads the live storefront, so content changes need no app update
  • Push tokens register to the customer record
  • Version gate that can require an update
  • You sign and submit the app to the stores yourself

Themes

Design

Design tokens control colours, type and spacing, and a business can offer more than one theme.

Colours, typography, spacing and component styles are defined as design tokens. Merchants adjust them in the admin theme editor, and a business can ship several themes with a switcher for shoppers where it is enabled. The theme copilot on the AI page can propose token changes from a plain-language request.

  • Design tokens for colour, type, spacing and components
  • Theme editor in the admin
  • Several themes per business, with a switcher where it is enabled
  • Theme copilot for plain-language changes

Content and SEO

Content

A news ticker, banners, a blog and product pages that carry structured data for search engines.

A news ticker shows a line of text with an optional link, and banners and hero blocks are edited in the admin. The blog produces Article, Organization and breadcrumb structured data, and product pages carry product structured data. FAQ structured data is available on product pages and on selected templates.

  • News ticker with text and a link
  • Banners and hero blocks edited in the admin
  • Blog with Article, Organization and breadcrumb structured data
  • Product structured data on product pages

Point of Sale

pos

A browser-based register for your counter, with cash-up, payment-machine support and a queue that keeps sales safe when the connection drops.

The QuantumOS POS runs in a browser or in a mobile shell on a tablet or phone. Cashiers sell by barcode or name, take cash, card, UPI or wallet payments, and close their shift with cash counted against what the system expects. Sales taken while the connection is down are queued on the device and sent when it comes back. Card machines connect through a payment hub that keeps card data on the machine, and each device is certified with you before it goes live.

A sales queue for when the connection drops

Resilience

Sales completed while the register is offline are saved on the device and sent to the server once it reconnects.

The register keeps a queue of completed sales in the browser's storage on the device. While the page stays open and the connection is down, the cashier can keep ringing up sales. When the connection returns, the register sends the queue and checks again every 30 seconds. Each sale carries its own local id, so a sale that is sent twice is stored once. Offline limits for a terminal (hours and number of sales) can be configured, but they are not enforced yet. The register does not keep a local copy of the whole catalog, so keep the page open rather than reloading it while you are offline.

  • Completed sales queue on the device and send automatically on reconnect
  • The queue is checked every 30 seconds while online
  • Each sale has a local id, so duplicates are not stored
  • Offline limits can be configured per terminal but are not enforced yet
  • Keep the register page open while offline; it does not cache the whole catalog

Payment machines

Payments

Connect card and UPI machines through one payment hub that never holds card data.

The payment hub gives every supported machine the same lifecycle: queued, waiting for the customer, processing, approved, declined, cancelled, expired or failed, plus a state for payments that need reconciling. A single card, UPI or wallet sale completes only after the machine approves it. Card data stays on the machine. The hub's compatibility roster covers Razorpay POS, Paytm POS, Pine Labs, PhonePe POS, BharatPe, Mswipe, Worldline, Innoviti, Ingenico and Verifone. The roster is a list of ecosystems we are built to work with, not a list of devices certified for you. Each machine is set up by an admin, tested on the real device, and only then activated. Provider credentials are stored encrypted and cannot be read back.

  • One payment lifecycle across the supported machine ecosystems
  • Card data stays on the payment machine
  • A sale completes only after the machine approves it
  • Duplicate device commands are prevented with tenant-scoped idempotency
  • Credentials stored encrypted and write-only
  • Machines are activated only after testing on the real device

Register basics

Barcode and name search, parked carts, discounts, customer lookup and price lists at the counter.

Cashiers find products by barcode or name, adjust quantities and apply item-level discounts. Sales, returns and exchanges are separate transaction types with automatic receipt numbers. A cart can be parked and recalled. Customers are found by name, phone or email. Voiding a line or editing a price asks for a manager PIN. Taxes are configured per location with HSN codes and calculated automatically, and price lists with rules can be applied.

  • Barcode scanners that act as a keyboard, plus name search
  • Sales, returns and exchanges with automatic receipt numbers
  • Parked carts and item-level discounts
  • Manager PIN to void a line or edit a price
  • Customer lookup by name, phone or email
  • Per-location tax rates with HSN codes, and price lists with rules

Shifts and cash-up

Open a shift with a float, close it with a count, and see the variance.

Each shift opens with a recorded opening balance. At close, the register shows the expected cash, the amount counted and the variance, and a Z-report summarises the shift. Each terminal has its own session, so a discrepancy points to a terminal rather than to a whole store. The cash drawer can open automatically on a cash sale.

  • Opening balance recorded for every shift
  • Expected cash, counted cash and variance at close
  • Z-report per shift
  • A session per terminal, so variances are traceable
  • Cash drawer opens on a cash sale when configured

Click and collect

Online orders can be collected in store, and the POS handles the handover.

Shoppers can choose to pick up an online order at a store. The store sees the order in a click and collect queue in the POS and hands it over. Ship-from-store routing for online orders is handled in the order management pages.

  • Pickup option at storefront checkout
  • Click and collect queue in the POS
  • Ship-from-store routing handled in order management

Receipts, scanners and counter hardware

Settings for receipt printers, barcode scanners, cash drawers and weighing scales.

The hardware settings cover receipt printing, barcode scanner input (including prefix and suffix handling), the cash drawer kick code and weighing scales. The Android shell prints receipts to thermal printers over ESC/POS. A simple kitchen display page shows orders as pending, preparing or ready. We do not publish a list of certified printers or scanners, so test your own hardware before you rely on it.

  • Receipt printer, scanner, cash drawer and scale settings
  • ESC/POS thermal receipt printing from the Android shell
  • Scanner prefix and suffix handling
  • Kitchen display page with pending, preparing and ready
  • No certified-device list is published; test your own hardware first

Cashier reports

Sales count, voids, revenue and average ticket for each cashier.

The POS reports give each cashier's number of sales, number of voids, revenue and average ticket for a period, alongside session and Z-report views for the store. Business customers can be invoiced from the counter with their GSTIN captured on the sale.

  • Sales, voids, revenue and average ticket per cashier
  • Session and Z-report views
  • GSTIN captured for business invoices at the counter

Orders, Inventory & Warehouse

ops

Reserve stock at checkout, route orders to the right location, and run receiving, picking and transfers from one system.

The ops layer connects orders to physical stock. Checkout reserves inventory before an order is created and refuses the order if the stock is not there. Orders are routed to a warehouse or store by pincode and by rules you set. The warehouse tools cover receiving, putaway, counts, transfers and wave picking on a mobile scanner screen, and carrier connectors create shipments with your own carrier accounts.

Order routing

Routing

Send each order to a warehouse, carrier or payment gateway using rules you write, and test the rules before they go live.

Routing rules are checked in priority order and the first match wins. A rule can choose a warehouse, a carrier, a payment gateway or a fulfilment priority, using fields such as order total, currency, channel, payment method, shipping state, city, pincode, country and customer tier. A simulator lets you try an order against your rules. For nearest-location fulfilment, a pincode lookup picks the location with the best delivery days and available stock.

  • Rules for warehouse, carrier, payment gateway and fulfilment priority
  • Conditions on order total, channel, payment method, location and customer tier
  • Priority order, first match wins
  • Simulator to test an order against your rules
  • Pincode lookup for the nearest location with stock

Stock reservation that fails closed

Checkout reserves stock before the order exists. If the stock is not there, no order or payment attempt is created.

When a shopper checks out, each product is reserved against the best available stock position in a single step, and the reservation is held for 60 minutes. If any product is short, the whole reservation is rolled back and no order is created. Products are locked in a fixed order so two carts with the same products cannot deadlock each other. Reservations are marked held, committed, released or expired, so you can see what happened to every hold.

  • Reserves stock before an order or payment attempt is created
  • A shortfall rolls the whole reservation back
  • Holds expire after 60 minutes
  • Held, committed, released and expired states are recorded
  • Row locking in a fixed order avoids deadlocks between carts

Receiving, putaway and counts

Warehouse

Receive goods, put them away, count stock and adjust it, with a movement record for each change.

Warehouses have zones and bins. Goods are received against receipts, put away to bins, counted in cycle counts and adjusted with a reason. Each stock movement is recorded. Warehouse staff use mobile scanner screens for picking, receiving, putaway, counts and dispatch, which run in a browser or in the WMS mobile app shell.

  • Zones and bins per warehouse
  • Receipts, putaway, cycle counts and adjustments
  • A movement record for each stock change
  • Mobile screens for pick, receive, putaway, count and dispatch

Wave picking

Group orders into pick waves ordered by urgency, with carrier cut-off times and priority taken into account.

Wave planning scores orders by how urgent they are, using carrier cut-off times and order priority, and splits them into waves of a fixed size. Pickers work through a wave on the mobile scanner screen. Wave planning does not group by zone or optimise the walking path yet.

  • Orders scored by carrier cut-off and priority
  • Waves of a fixed size, most urgent first
  • Picking on the mobile scanner screen
  • No zone grouping or path optimisation yet

Transfers and store fulfilment

Move stock between locations with an approval trail, and fulfil online orders from stores.

A stock transfer goes through requested, approved, in transit and received, so there is a record of who moved what and when. Online orders can be picked up in store with click and collect, shipped from a store, or placed by store staff from a store's own screen when an item is not stocked there.

  • Transfers move through requested, approved, in transit and received
  • Click and collect at a store
  • Ship-from-store routing
  • Endless aisle: staff place an order for stock held elsewhere

Returns and store credit

Customers request a return from their order and choose a refund, a replacement or store credit.

A returns request from the storefront records the reason and the outcome the customer wants: a refund, a replacement or store credit. Store credit is added to a wallet whose ledger is append-only: each credit and debit is a new row, and credits can carry an expiry date. There is no guided grading workflow for returned items yet.

  • Return requests from the customer's order
  • Refund, replacement or store credit as the outcome
  • Append-only wallet ledger for store credit
  • Optional expiry on credits
  • Guided grading of returned stock is not available yet

Lot, serial and expiry

Lot and serial numbers travel with receipts, picks and movements, and expiring batches can be looked up.

Lot and serial identifiers are recorded when goods are received and carried through picks and movements, so you can see where a batch went. Batches and serial numbers can be browsed, and an expiring-batch lookup shows stock approaching its expiry date. Picking does not yet enforce first-expiry-first-out or reject a wrong lot automatically.

  • Lot and serial ids on receipts, picks and movements
  • Batch and serial browsing
  • Expiring-batch lookup
  • First-expiry-first-out is not enforced yet

Carrier connectors

Create shipments, labels and tracking with your own carrier accounts.

Connectors for UPS, FedEx, DHL, USPS, Royal Mail and Australia Post use your own carrier credentials, and Indian carriers including Delhivery, Blue Dart, DTDC and Shiprocket are used for domestic shipping. The connectors create shipments and labels and fetch tracking, and some can quote rates; what each carrier's API offers differs. Order routing does not yet fetch live carrier rates to choose between carriers.

  • Connectors for UPS, FedEx, DHL, USPS, Royal Mail and Australia Post
  • Indian carriers including Delhivery, Blue Dart, DTDC and Shiprocket
  • Your own carrier account credentials
  • Rate quotes only where the carrier's API provides them
  • Routing does not yet choose a carrier from live rates

Marketplaces & Channels

marketplace

Run your own multi-vendor marketplace, and send your catalog to Google, Meta, Amazon, Flipkart and other channels.

This pillar covers selling through other people and on other channels. You can open a multi-vendor marketplace with seller onboarding, listing approval, commissions, settlements and disputes. You can publish product feeds to advertising and shopping channels, pull Amazon and Flipkart orders into your order system, and watch competitor prices to reprice. Some channels are outbound feeds only, and a few connectors are still planned, as each section says.

Multi-vendor marketplace

Marketplace

Onboard sellers, check their documents, approve their listings and let them sell on your storefront.

Sellers apply with company details and KYC documents, which your team approves or rejects. Bank details can be verified, brand authorisations are recorded, and restricted categories and risk signals are tracked. Sellers submit listings, and your team approves them or rejects them with a reason. Buy-box logic decides which seller's offer is shown when several sell the same product, and sellers receive ratings.

  • Seller applications with KYC documents approved by your team
  • Bank verification and brand authorisations
  • Restricted categories and risk signals
  • Listing approval with a reason on rejection
  • Buy-box and seller ratings

Commissions and settlements

Set commission rules, settle sellers on delivered orders, and account for tax collected at source and deducted at source.

Commission rules apply by category, brand, seller, listing, seller tier or across the marketplace, and the rule with the highest priority wins. Settlements are built from delivered orders only and show the commission, GST on the commission, tax collected at source, tax deducted at source and any reserve held back. The default settlement cycle is seven days. Marking a settlement as paid is a manual step in the admin, with an optional payment reference, and seller invoices can be generated from a settlement.

  • Commission rules by category, brand, seller, listing, tier or everything, with priorities
  • Settlements built from delivered orders only
  • GST on commission, TCS and TDS shown on each settlement
  • Default seven-day settlement cycle
  • Payouts are marked paid by hand; there is no built-in payout rail or escrow

Disputes and returns

Sellers and buyers can raise disputes and returns, and your team works them from the admin.

Disputes and marketplace returns are recorded against the order and the seller, and your team resolves them from the admin. Outcomes feed the seller's record. The platform does not enforce a response timer or automatic refunds for disputes today.

  • Disputes and returns recorded against order and seller
  • Worked from the admin
  • No response timer or automatic refunds yet

Product feeds to channels

Channels

Publish your catalog as feeds for Google Shopping, Meta, Instagram, TikTok, Pinterest, Bing, Snapchat and Meesho.

Feeds are generated per channel from your catalog with the field mapping each channel expects. Most channels take your feed as an outbound file: Google Shopping, Meta, Instagram, TikTok, Pinterest, Bing and Snapchat, plus a CSV for Meesho. This is a feed, not a two-way connection, so listings and inventory on those channels are updated when they read the feed. eBay and Etsy connectors are planned and are not available today.

  • Feeds for Google Shopping, Meta, Instagram, TikTok, Pinterest, Bing and Snapchat
  • CSV feed for Meesho
  • Field mapping per channel
  • Outbound feeds, not two-way sync
  • eBay and Etsy are planned, not available

Amazon and Flipkart orders

Pull orders from Amazon and Flipkart into your order system on a regular poll.

Connectors for Amazon and Flipkart fetch new orders every 15 minutes and bring them into the same order pipeline as your own orders. Orders arrive on that poll rather than instantly. Inventory is not pushed back to those marketplaces by this connector.

  • Amazon and Flipkart orders pulled in on a 15-minute poll
  • Orders enter the same pipeline as your own
  • Not instant, and no inventory push back to the marketplace

Competitor price monitoring and repricing

Watch competitor prices on the sites you choose and let the repricing regent propose price changes, with price history and revert.

You keep a registry of competitors and the channels to watch: their own site, Amazon, Flipkart, Google Shopping or Meta. A scraping job collects their prices, and the repricing regent evaluates your rules every six hours. Applying a decision writes the new price, records it in a price history with the time it was applied, and can be reverted. A backtest job shows how a rule would have behaved, and a kill switch stops the regent.

  • Competitor registry across site, Amazon, Flipkart, Google Shopping and Meta
  • Price collection and rule evaluation on a six-hour cycle
  • Price history for every applied change, with revert
  • Backtest job and kill switch
  • Prices are scraped from public pages, not read from marketplace APIs

Ad spend and attribution

Bring Google and Meta ad spend next to orders and compare attribution models.

When ad sync is enabled, spend from connected Google and Meta ad accounts is read into the platform every 30 minutes. Orders can be attributed to marketing touches using first touch, last touch, linear, time decay or position-based models. TikTok and Pinterest spend are not imported, and there is no incrementality or data-driven attribution model today.

  • Google and Meta ad spend read in when ad sync is enabled
  • First touch, last touch, linear, time decay and position-based models
  • TikTok and Pinterest spend are not imported

Listing health score

A simple score for each marketplace listing based on its images, description, attributes, HSN code and rating.

Each listing gets a score from five checks: whether it has images, how complete its description is, how many attributes are filled, whether it has an HSN code and what its rating is. It helps you spot weak listings. There is no automatic fix queue yet.

  • Scored on images, description, attributes, HSN code and rating
  • Helps you spot weak listings
  • No automatic fix queue yet

Loyalty, CRM & Marketing

loyalty

Points, tiers and referrals, customer segments, and message journeys, all run on your own customer and order data.

Loyalty, CRM and marketing share one customer record. Customers earn points on their orders, move through tiers, refer friends and redeem rewards. You build segments from customer and order fields, and use them in journeys that send SMS, WhatsApp or email through the provider you already use. The sections below say what each part does today and where its limits are.

Points, tiers and rewards

Customers earn points on orders, move through tiers and redeem rewards, with bonus campaigns for promotions.

A purchase rule sets how many points a customer earns per 100 of order value. The earn is multiplied by the customer's tier and by any active bonus campaign, and the result is written to a ledger. The rules also support points for reviews, sign-ups, referrals, shares, birthdays and streaks. Tiers, a rewards catalog, missions and badges sit on top. A return reverses the points earned on that order, in full. Points can expire, and a liability view shows what you owe in points.

  • Points per 100 of order value, with a multiplier for each tier
  • Time-boxed bonus campaigns with their own multiplier
  • Other earn actions: reviews, sign-ups, referrals, shares, birthdays and streaks
  • Tiers, rewards catalog, missions and badges
  • A return reverses the whole earn for that order
  • Point expiry and a liability view

Referrals and affiliates

Reward customers who refer friends, and pay affiliates who bring in sales.

Referrals reward both sides: the referrer gets a reward and the new customer gets a first-order benefit. A separate earn programme supports affiliates with tiers, attribution windows, conversion tracking and payout batches, with payouts by UPI, bank transfer or credit and tax deducted at source calculated on each batch. The affiliate programme runs separately inside each business, not as a shared pool across brands.

  • Two-sided referral rewards
  • Affiliate tiers, attribution windows and conversion tracking
  • Payout batches by UPI, bank transfer or credit, with tax deducted at source
  • Runs inside each business, not as a cross-brand pool

Customer segments

Build segments from customer and order fields with nested AND and OR rules, and preview the size before you save.

A segment is a set of rules joined by AND and OR, with groups nested inside groups. Rules use customer fields: email, name, phone, city, country, sign-up date, order count, lifetime value, last order date, points, lifetime points and tier. The segment is evaluated against your live data when you ask for it, so it is as current as the data, but it is not a continuously updated stream. It does not yet use browsing session, return or support-ticket events.

  • Nested AND/OR groups
  • Fields: order count, lifetime value, last order date, points, tier, city, country and more
  • Evaluated on request against live data
  • Count and sample preview
  • Session, return and support-ticket events are not available as fields yet

Journeys

Automation

A visual canvas for message flows with triggers, waits, conditions and actions.

Journeys are built on a canvas from trigger, condition, action, wait and end nodes. A condition can split customers into branches, including an A/B split. Actions send an SMS, WhatsApp or email message or move the customer to a new journey stage. Steps for adding a tag, adding to a segment or applying a discount are recorded as having run, but they do not yet change the customer or create a discount. Each customer's progress through a journey is logged node by node. Journeys are processed on a 10-minute cycle, so a message goes out within about that window of its trigger rather than instantly. There is no dry-run mode.

  • Trigger, condition, action, wait and end nodes
  • Condition branches, including A/B splits
  • Actions that send a message or update the customer's journey stage
  • Tag, segment and discount steps are only recorded for now
  • A per-node run log for each customer
  • Processed on a 10-minute cycle

Messages through your own providers

Send SMS, WhatsApp and email through the providers you already use.

Journey actions send through the provider configured for your business. SMS goes through Exotel or MSG91, WhatsApp through MSG91, Gupshup or Interakt, and email through SendGrid, Brevo or MSG91. A Mailgrid email client is also available. Push notifications are not part of the journey sender today.

  • SMS: Exotel and MSG91
  • WhatsApp: MSG91, Gupshup and Interakt
  • Email: SendGrid, Brevo and MSG91
  • Push notifications are not sent from journeys

Back-in-stock alerts

Tell customers when a product they wanted is available again.

A scheduled job checks products that customers asked to be told about and sends the alert when stock returns. It runs alongside the other lifecycle messages.

  • Customers subscribe to a product they want
  • A scheduled job sends the alert when stock returns

Attribution

Credit orders to marketing touches with first-touch, last-touch, linear, time-decay or position-based models.

Orders can be attributed to the marketing touches that led to them using five rule-based models: first touch, last touch, linear, time decay and position based. These are standard multi-touch models, not causal measurements, so use them to compare channels rather than to prove what caused a sale.

  • Five rule-based multi-touch models
  • Compare channels under different models
  • Not a causal or incrementality measurement

Platform & Developer Experience

platform

A scoped REST API, OAuth, signed webhooks, an MCP server and a developer portal for building on your store data.

The platform exposes your catalog, orders, inventory and customers to your own code. Developers get scoped API keys or OAuth tokens, a REST API with consistent error responses, signed webhooks with a delivery log, and an MCP server for AI tools, all managed from a developer portal in the admin. Behind the scenes, feature flags, tenant snapshots and per-tenant backup encryption support safer releases and recovery.

REST API with scoped access

A versioned REST API for products, orders, inventory, customers, feeds and webhooks, limited by scopes you grant.

The public API lives under /api/v1 and covers products, orders, inventory, customers, feeds, webhooks and the calling key's own details. Every request is limited by scopes such as products:read, orders:write or inventory:write, so a key can do only what you gave it. Errors come back in a standard problem-details format, and responses include rate-limit headers. The rate limiter keeps its counts in memory on each server instance, so treat the headers as guidance rather than an exact quota.

  • Resources: products, orders, inventory, customers, feeds and webhooks
  • Twelve scopes, read and write, enforced on every call
  • Standard problem-details error responses
  • Rate-limit headers on responses
  • The rate limiter counts per server instance, so limits are approximate

OAuth and API keys

Issue API keys or let apps connect with OAuth 2.0, and revoke or rotate them from the portal.

Apps can connect with OAuth 2.0 using the authorisation-code flow with PKCE, refresh tokens or client credentials. Access tokens last one hour and refresh tokens last 30 days. API keys are issued from the developer portal and can be rotated there.

  • Authorisation code with PKCE, refresh tokens and client credentials
  • One-hour access tokens and 30-day refresh tokens
  • API keys that can be rotated from the portal

Developer portal

Manage keys, OAuth apps, webhooks and usage, and read the docs, from the admin.

The developer portal in the admin has pages for API keys, OAuth apps, webhooks, usage and documentation. The usage page shows API request volume and error rates. There is no interactive request builder yet.

  • Keys, OAuth apps, webhooks, usage and docs in one place
  • API request volume and error rates on the usage page
  • No interactive request builder yet

OpenAPI document

A maintained OpenAPI 3.1 description of the public API.

An OpenAPI 3.1 document is served at /api/openapi.json. It is maintained by hand, covers the public API as a whole rather than per business, and is cached for an hour. It is not generated from the code on each release, so check it against the API when you depend on a detail.

  • OpenAPI 3.1 at /api/openapi.json
  • Maintained by hand, one document for the whole API
  • Cached for an hour

MCP server for AI tools

New

Let an AI tool search your catalog, read products and create orders through a scoped MCP connection.

A tenant-scoped MCP server speaks JSON-RPC and exposes six tools: search products, get a product, get recommendations, list categories, create an order and update inventory. The caller needs the mcp:invoke scope plus the scope for any write. It does not expose customers or segments.

  • Six tools: search products, get product, recommendations, categories, create order and update inventory
  • Scoped by the key or token that connects
  • Customers and segments are not exposed

TypeScript client

A small TypeScript client for the REST API.

The @quantum/sdk package is a dependency-free TypeScript client written by hand for the REST API. It is at an early version, and there are no official Python, Go or Ruby clients. Use the REST API and the OpenAPI document for other languages.

  • Dependency-free TypeScript client
  • Early version (0.1.0), written by hand
  • No official Python, Go or Ruby clients

Signed webhooks

Subscribe to events and receive HMAC-signed deliveries, with a log of every attempt.

Webhook deliveries are signed with HMAC-SHA256 so your server can verify them. Each attempt is logged with its status and response. A delivery is tried three times with a growing delay, and one that still fails goes to a dead-letter list that is retried after ten minutes. You can send a test event from the admin. There is no one-click replay of old events and no secret-rotation overlap yet.

  • HMAC-SHA256 signatures
  • A log of every delivery attempt
  • Three attempts, then a dead-letter list retried after ten minutes
  • Test events from the admin
  • No replay of past events and no secret rotation overlap yet

Feature flags

Turn features on for a percentage of visitors, or switch them off with a kill switch.

Flags can be set for one business or for everyone, with a deterministic percentage rollout so a visitor stays in the same group, and a kill switch. Changes are written to an audit table. Flags are cached for about a minute, so a kill switch can take that long to reach every server.

  • Per-business and global flags
  • Deterministic percentage rollout
  • Kill switch
  • Audit table of changes
  • Cached for about a minute, so changes are not instant

Tenant snapshots

Take point-in-time snapshots of a business's database and restore from them.

Snapshots of a business's database are taken using Neon branches, which are copy-on-write, and recorded in a snapshot registry that the admin shows. You can restore from a snapshot. This is a snapshot and restore tool, not a writable sandbox that is anonymised and exposed on its own address.

  • Point-in-time snapshots using Neon branches
  • Snapshot registry in the admin
  • Restore from a snapshot
  • Not a writable, anonymised sandbox

Encrypted tenant backups

Each business's backup is encrypted with its own key.

Backups are encrypted per business, so removing that business's key makes its backup unreadable. Audit logs follow a PII-redaction policy, and data-subject request types are defined for handling access and erasure requests. This is a set of controls to discuss with your security reviewer, not a certification.

  • A separate encryption key for each business's backups
  • Deleting the key makes the backup unreadable
  • PII redaction policy for audit logs
  • Data-subject request types defined

SOL ORBIT Cache Architecture

sol-orbit

Sun, Origin, and Luna: a cache control plane for edge delivery, tenant truth and freshness.

SOL ORBIT is the QuantumOS X3 speed architecture. Sun keeps public storefront traffic bright at the edge. Origin is Earth: the grounded layer where tenant truth and read models stay correct. Luna is Moon: the freshness layer that moves invalidation, prewarm, and observability whenever truth changes. The model is backed by a control plane in the admin: 100 controls, tenant policy documents, approval-gated changes, a freshness ledger, route headers, and scorecard records. Not every control is bound to a live action yet; each is marked as an API binding, an approval gate, a policy definition or a runbook.

Cache control plane

Control plane

SOL ORBIT is a persisted operator console with policies, overrides, approvals and a freshness ledger for each tenant.

The platform stores cache policies, per-control overrides, approval/action requests, freshness ledger rows, and scorecards. Super-admins can seed all tenants by industry, tune rollout percentage, save safe controls immediately, and route risky changes through two-person approval before they touch live traffic.

  • 100 controls mapped across Sun, Origin/Earth, Luna/Moon, Orbit, and Eclipse; 39 are API bindings, 18 approval gates, 35 policy definitions and 8 runbooks
  • Tenant policy documents with speed profile, industry profile, rollout, route contracts, and safeguards
  • Approval queue for dangerous cache actions and private-route changes
  • Freshness ledger plus scorecard records for hit rate, p95, Core Web Vitals, origin pressure, invalidation lag, and approval debt
  • Response headers: X-SOL-Layer, X-SOL-Profile, and X-SOL-Policy

Sun edge delivery

S

Public HTML, JSON, search, images, and assets live as close to the shopper as the stack can safely place them.

Sun combines OpenNext KV ISR, Cloudflare Cache API hot lanes, response cache headers, immutable assets, image transformations, and per-brand static route chunks. The rule is strict: only tenant-safe, anonymous, variant-keyed content belongs here.

  • OpenNext incremental cache and tag cache
  • Per-PoP Cache API for hot anonymous search and storefront APIs
  • Cache-Control, Cloudflare-CDN-Cache-Control, Cache-Tag, X-Cache
  • Image transformation and owned CDN asset delivery
  • Per-brand route chunks to avoid shipping every tenant home to every tenant

Origin / Earth truth layer

O

Origin is Earth: it keeps tenant, catalog, price, stock, CMS, order, and search truth durable and isolated.

Origin is not a public cache. It is the Earth layer: authoritative truth plus prepared read models such as Hyperdrive/Postgres, D1 tenant planes, product/search matviews, native KV read-through, catalog SWR, home snapshots, and root tenant maps.

  • Hyperdrive/Postgres and per-tenant D1/Neon/Supabase data planes
  • Product/search materialized views for list, PDP, and category reads
  • Native KV read-through for theme, routing, CMS, catalog, and nav
  • Home and catalog snapshots with stale-good fallbacks
  • Explicit tenant/theme root maps for Worker-safe resolution

Luna / Moon freshness control

L

Luna is Moon: it keeps fast content honest with cache tags, exact deletes, prewarm, version bumps, and gauges.

Luna moves whenever Origin changes. Product edits, category changes, content publishes, theme changes, search configuration edits, and domain updates all map to bounded invalidation actions and prewarm routes.

  • Shared CACHE_TAGS registry for publish events
  • /api/x3/cache/invalidate for CF tags, KV prefixes, and exact PDP keys
  • Home snapshot version bumps and route prewarm
  • Server-Timing, X-Cache, and image verification gauges
  • Eclipse Bypass for checkout, account, sessions, and mutations

100-system feature scan

Research

The architecture borrows public cache patterns from commerce platforms, CDNs, framework hosts, search systems, media CDNs, and database caches.

The scan is an inventory of public documentation, not a benchmark, and it makes no claim that X3 is faster than any of those systems. It maps full-page cache, subrequest cache, surrogate/cache tags, SWR/SIE, origin shield, durable edge cache, query cache, image transformation cache, search-session cache, negative-result cache, stampede locks, and prewarm into SOL ORBIT-native names and guardrails.

  • Full-page cache patterns adapted to X3 tenant isolation
  • Tag invalidation mapped to the CACHE_TAGS registry and tenant scopes
  • Tiered and origin-shield patterns planned for campaign traffic (not built yet)
  • Search caching for anonymous query hot lanes
  • Hyperdrive and materialized-view read paths for database-heavy catalogs

Tenant cache profiles

Profiles

Industry defaults tune cache lifetimes to each kind of business's freshness risk and traffic shape.

Electronics needs aggressive PDP/search/image cache with precise stock and price invalidation. Food needs short TTLs for slots and serviceability. Professional services needs hot service pages and no-store lead workflows. Travel needs seasonal prewarm and D1/R2 media planes. Marketplace/B2B needs vendor and buyer-group variants.

  • Electronics, imaging, and dense PIM catalogs
  • Grocery, beverages, restaurants, and D2C foods
  • Finance, healthcare, spa, and other service sites
  • Destinations, packages, journals, and campaigns
  • 16 industry profiles in the admin

ORBIT operating loop

Ops

Observe, Route, Build, Invalidate, Tune turns cache from a header trick into an operating system.

The loop makes cache reviewable. Every public route declares its layer, variants, invalidation events, and observability markers. Every publish should say what Luna moved. The aim is for every tenant to show hit rate, origin trips, invalidation lag, prewarm lag, image failure rate, and p95 TTFB; today a benchmark probe and scorecard records cover part of that.

  • Observe cache state with Server-Timing, X-Cache, RUM, and image checks
  • Route public traffic to Sun and private traffic to Eclipse Bypass
  • Build narrow Origin/Earth read models before demand
  • Invalidate via exact tags, keys, prefixes, and version bumps
  • Tune TTLs, prewarm routes, and campaign mode by tenant profile

AI & Intelligence

ai

AI that works from your orders, catalog and customers, with your own model keys and rules you control.

The AI capabilities in this pillar read from your own catalog, orders and customers rather than generic text. Search, recommendations and fraud screening are specific to your business because they run against your records. Kynetra AI is the layer behind the autonomous parts: Kynetra Prime proposes search improvements, and separate regents handle repricing, demand and ad accounts. You choose which models run by registering your own provider keys, and you set budgets, guardrail rules and approval steps.

Kynetra Search

New

A search index kept in your own database and updated whenever a product is saved, with fuzzy matching, synonyms, redirects and ranking you can tune.

Kynetra Search reads from a flattened product search table in your own database instead of joining raw catalog tables on every query. The database keeps that table current automatically, so saving a product updates its search row in the same step and no scheduled rebuild is needed. Queries are cleaned up, matched with fuzzy text matching, expanded with your synonyms and ranked by weighted signals: exact name matches, prefixes, matches on every word, brand, category, whether the product can be bought, and whether it has an image. In Search Studio, merchandisers pin, boost or bury products, set redirects and zero-result fallbacks, and adjust the ranking weights per business.

  • Search table updated automatically when a product is saved, with no scheduled rebuild
  • Fuzzy text matching so small typing errors still find the product
  • Synonyms, redirects and pinned, boosted or buried products managed in Search Studio
  • Zero-result fallbacks: choose the products to show for a query that finds nothing
  • Ranking weights for exact, prefix, all-word, meaning-based, in-stock and image signals, set per business
  • Last good results are served if the database errors, and identical queries at the same moment share one lookup

Kynetra Prime — search auto-training

New

Prime watches customer search activity and proposes relevance changes. Risky changes wait in a review queue until you approve them.

Kynetra Prime watches search activity such as queries that return nothing, clicks and add-to-carts from search, and proposes improvements: synonyms, redirects, query rewrites, zero-result recoveries and boosts. A discovery run also asks a language model to look at recent search events and suggest more. The engine applies only low-risk changes by itself. Anything riskier appears in a review queue, and it goes live only when an admin approves it. A kill switch and a percentage rollout control whether the feature runs at all, and a quality dashboard shows zero-result rate, click-through and training history.

  • Learns from search events: queries with no results, clicks and add-to-carts from search
  • Proposes synonyms, redirects, query rewrites, zero-result recoveries and boosts
  • Risky changes wait in a review queue and go live only when an admin approves them
  • Low-risk changes are applied automatically, so review the training history
  • Kill switch and percentage rollout for the feature flags that gate it
  • Quality dashboard with zero-result rate, click-through and training runs

Meaning-based search and related products

Core

Vector search on the storefront finds relevant products when keywords match nothing, and powers related-product suggestions.

On the storefront, the shopper's words are turned into an embedding with a Cloudflare Workers AI model and matched to the nearest products in a vector index kept separately for each business. This finds relevant products even when no keyword matches. Repeated queries reuse a cached embedding. The lookup runs against a time limit behind a circuit breaker, and if it is slow or unavailable, shoppers get the keyword results instead. The same index powers a more-like-this suggestion for a product. Meaning-based fallback can be switched off per business in the search settings. Image search needs a vision and vector service to be configured for the business.

  • Embeddings from a Cloudflare Workers AI model, with a separate vector index for each business
  • Finds products by meaning when keyword search finds little or nothing
  • More-like-this suggestions from the same index
  • Query embeddings cached for repeat searches
  • Time limit and circuit breaker: shoppers fall back to keyword results if the lookup is down
  • Meaning-based fallback switchable per business in search settings

Command Chat — the admin copilot

Core

Ask questions about orders, catalog and analytics in plain language. It reads live data through the admin's own tools, as you, and cannot change anything yet.

Command Chat is the operations copilot inside the admin. It acts as the signed-in user on their business, so it can only reach what that user's role allows. It answers by calling the admin's own read tools, shows each tool call as it happens, and links the records it mentions. In this release it is read-only: if you ask for a change, it says so and points you to the screen where you can make it. Each tool call is recorded with the session, user, procedure, arguments, result status and time.

  • Works as the signed-in user, limited to that user's role and business
  • Calls the admin's own read tools, and shows each call in the conversation
  • Read-only in this release; write actions are not available yet
  • Each tool call is logged with session, user, procedure, arguments, result status and time
  • Uses your model keys through the same AI gateway as every other AI feature

Theme copilot

Core

Describe storefront theme changes in plain language and see them applied to the live preview, with undo.

In the theme editor, you describe a change such as a transparent header or bigger, rounder buttons. The copilot sees your current theme, returns a checked set of design-token changes with an explanation, and layers them over the theme in the editor. Because the edit goes through the editor's normal history, you can undo it, preview it and let autosave keep it. It works across several turns, so each request builds on what is already there.

  • Reads the current theme, so each request is relative to what is there now
  • Returns design-token changes that are checked before they are applied
  • Changes flow through the editor's undo, redo, preview and autosave
  • Works over several turns in one conversation
  • Uses your model keys through the AI gateway

Conversation engine

A flow-based bot engine behind the WhatsApp and Telegram assistants. Language-model intent detection is built but not connected yet.

Conversations run on a session store that remembers where each customer is in a flow. A flow can search the catalog, add items to a cart, create a payment link and flag a conversation for a person to take over. A separate agent that reads a message and returns an intent (browse, add to cart, checkout, payment, complaint, return, negotiate or greeting), the items and quantities mentioned and whether to escalate has been built, but it is not part of the live flow yet, so today's bots follow the flows you design. See the conversational commerce page for the channels that are live.

  • Sessions remember the customer's place in the flow
  • Flows for catalog search, add to cart and payment links
  • A handover flag for a person to take over
  • Intent detection agent built, not yet connected to the live bot
  • Used by the WhatsApp and Telegram channels

Fraud screening

Rules with weights score each new order and place it in one of four outcomes: allow, review, hold or block.

After an order is created, active fraud rules are checked against it and the weights of the rules that match are added up. A score under 30 is allowed, 30 to 60 goes to review, 60 to 90 is held, and above 90 is blocked; a rule set to block blocks the order on its own. Rules you can use include order velocity, large first orders, brand-new accounts paying with a chosen method, email address patterns, country mismatches and a blocked-payment-methods list. Held and blocked orders are put on hold with the reason, and the admin shows every triggered rule with its weight so a person can decide. The score and the rules that fired are saved with the order.

  • Rule types: order velocity, order amount, account age, email patterns, country mismatch and blocked payment methods
  • Four outcomes: allow, review, hold and block
  • Each rule has a weight and an action you set
  • Held and blocked orders go on hold with a reason
  • The score and the rules that fired are saved with the order
  • Review queue and rule editor in the admin

Segmentation

Build customer segments from nested AND/OR rules on lifetime value, order history, tier, points and location.

A segment is a set of rules joined by AND and OR, and groups of rules can be nested inside each other. Rules run against your live customer and order data when you ask for the segment, with no copy to a separate analytics store. Fields include email, name, phone, city, country, sign-up date, order count, lifetime value, last order date, loyalty points and loyalty tier. Segments can be used in loyalty and marketing journeys.

  • Nested AND/OR rule groups
  • Fields: order count, lifetime value, last order date, points, tier, city, country and sign-up date
  • Rules run against your live customer and order data, with no separate analytics store
  • Used by marketing journeys and campaigns in the admin
  • Operators include equals, greater or less than, in, contains, starts with and is empty
  • A count and sample preview before you save a segment

Marketing agent

Recommends coupons, campaigns, segments, journeys, bundles and price changes, with budget and margin guardrails. You approve each one or let safe ones apply.

The marketing agent looks at your orders, stock and customers and fills an inbox of recommendations. Triggers include ageing inventory, lapsed customers, abandoned carts, festivals, new-customer welcomes, replenishment and weekend flash sales. A recommendation can be a coupon, promotion, campaign, segment, journey, price change, bundle or referral. You can approve, reject, edit or schedule each one. Guardrails cap discounts per day and per month and protect a minimum margin, and the agent can be set to apply recommendations itself within those limits.

  • Recommendation kinds: coupon, promotion, campaign, segment, journey, price change, bundle, referral
  • Triggers: ageing inventory, lapsed customers, abandoned carts, festivals, replenishment, weekend flash sales
  • Approve, reject, edit or schedule each recommendation, singly or in bulk
  • Daily and monthly discount budgets and a minimum margin as guardrails
  • Optional auto-apply within those limits; approval stays manual unless you turn it on
  • Approvals and applies are limited by role permissions

Description enhancer

Generate or improve product descriptions in bulk with Claude, GPT or Gemini, and keep the drafts for review.

Choose products and a model, and the enhancer writes descriptions in the tone you pick, using the product's own details and, if you allow it, the brand's website or Google Shopping snippets. Bulk runs are background jobs with progress and per-item status, so a large batch does not block the page. You can run it in draft mode and review each result before replacing the live description. Each model shows a credit cost in the settings, and requests are recorded in the AI gateway with the model, feature and token use.

  • Models: Claude, GPT and Gemini
  • Tone choices such as professional, conversational, technical or minimalist
  • Draft mode keeps results for review before they replace live descriptions
  • Bulk runs as background jobs with per-item status
  • Optional sources: the brand's website or Google Shopping snippets
  • Every request is recorded in the AI gateway log with model, feature and token use

Your own keys and guardrails

Register your own provider keys, set budgets, and block personal data, risky prompts and expensive calls with rules you choose.

You can add your own provider keys in the admin. They are encrypted at rest with AES-256-GCM and only the last four characters are shown again. Calls use your key when it matches the model's provider. When a model fails, the gateway tries the fallback models you have configured. Guardrail rules run on prompts and responses: block or redact personal data patterns such as email, phone, card number, Aadhaar and PAN, flag prompt-injection patterns, set a maximum cost per call, allow or deny models, block output patterns and rate-limit calls. Spend budgets, set by scope and period, block calls once they are exceeded, and alerts fire as usage approaches the limit. Usage on your own key is billed by the provider; AI usage on the platform's shared key is billed separately from the subscription, as the pricing page explains.

  • Provider keys encrypted at rest with AES-256-GCM; only the last four characters are shown again
  • Uses your key when it matches the model's provider
  • Fallback models if the first choice fails
  • Guardrail rules: personal data, prompt injection, cost per call, model allow and deny lists, output patterns and rate limits
  • Spend budgets by scope and period that block calls once exceeded, with usage alerts
  • Rules can apply to chosen features only, and a test panel tries a prompt against your rules

Search resilience

Architecture

Time limits, circuit breakers and cached fallbacks keep search answering when a dependency is slow or failing.

Slower steps such as meaning-based lookups and AI calls run against a time limit, so one slow dependency cannot hang a search. A circuit breaker stops calling a dependency that has failed repeatedly, then lets one trial request through to see whether it has recovered. Performance budgets are stored per metric, and a budget breach can switch an expensive step off so search degrades gracefully. If the database errors, the last good result for that search is served from cache. Identical queries arriving together are answered from one lookup. Individual features also sit behind kill switches.

  • Time limit on slower search steps
  • Circuit breaker that stops calling a failing dependency, then retries with one trial request
  • Stored performance budgets that can switch an expensive step off
  • Last good results served from cache if the database errors
  • Identical simultaneous queries share one lookup
  • Kill switches for individual search features

Conversational Commerce

conversational

Chat with customers on WhatsApp and Telegram: browse the catalog, add to a cart, get a payment link and be handed to a person.

A single bot runtime handles WhatsApp and Telegram conversations. It keeps a session for each customer, runs a flow you design to search the catalog, add items to a cart and send a payment link, and can flag a conversation for a person. WhatsApp order updates go out through the provider you already use. Language-model intent detection and the Conversation OS plugin family are not available yet, as the sections below say.

WhatsApp and Telegram bot runtime

One orchestrator handles inbound messages from WhatsApp providers and Telegram, and replies on the same channel.

Inbound messages from WATI, Gupshup, AiSensy and MSG91 on WhatsApp, and from Telegram, are normalised into the same shape and handled by one orchestrator. Each customer has a session stored in the database that remembers where they are in the flow, and messages are de-duplicated so a repeated webhook does not trigger a second reply. Telegram supports inline keyboards and verifies a secret token on the webhook.

  • WhatsApp through WATI, Gupshup, AiSensy or MSG91, and Telegram
  • One runtime and one session store for both channels
  • Repeated webhooks are de-duplicated
  • Telegram inline keyboards and webhook secret verification

Flows for catalog, cart and payment link

Design a flow that searches the catalog, adds items to a cart and sends a payment link.

Conversations run as a graph of nodes. Nodes can search the catalog, add items to a cart and create a payment link through the first payment provider configured for your business. A node can also flag the conversation for a person to take over. Flows are matched on what the customer sends, not understood in free language. There is no promo-code entry, delivery-slot selection or invoice sending in the chat today.

  • Node-graph flows
  • Catalog search and add to cart
  • Payment link from the first configured provider
  • A handover flag for a person to take over
  • No promo codes, delivery slots or invoices in chat yet

WhatsApp order updates

Send order confirmed, shipped, delivered, cancelled and refunded messages through your WhatsApp provider.

When an order changes state, a template message goes out to the customer through the WhatsApp provider connected to your business. The states covered are confirmed, shipped, delivered, cancelled and refunded.

  • Confirmed, shipped, delivered, cancelled and refunded messages
  • Sent through the provider you connect

Which business a chat belongs to

Incoming chats are matched to a business through its provider settings.

For WhatsApp, the sender number, project or instance on the provider's webhook is matched against the settings saved for each business. Telegram does not include the bot token in its updates, so it is matched to the single active Telegram integration. A deployment where several businesses share one Telegram setup is not fully separated yet, so run one Telegram bot per deployment for now.

  • WhatsApp matched by the provider's sender number, project or instance
  • Telegram matched to the active Telegram integration
  • Several businesses on one shared Telegram setup are not fully separated yet

Language-model intent detection

Roadmap

An agent that classifies what a customer wants and pulls out items and quantities exists, but is not connected to the live bot yet.

A conversational-commerce agent has been built that reads a message and returns an intent (browse, add to cart, checkout, payment, complaint, return, negotiate, greeting or other), any items and quantities, suggested actions, and whether a person should take over. It is not yet part of the live WhatsApp and Telegram flow, so today's bots follow the flows you design rather than interpreting free text.

  • Built: intent, items and quantities, suggested actions and escalation flag
  • Not yet connected to the live bot flow
  • Today's bots follow designed flows

Conversation OS plugin family

Roadmap

Four plugins for WhatsApp AI agents, marketing automation, a sales CRM and an AI call receptionist are catalogued, and are not available to run yet.

The marketplace catalogue defines four Conversation OS plugins: WhatsApp AI Agents, Conversation Marketing Automation, Conversation Sales CRM and AI Call Receptionist. Each has a description, channels and default limits, and the admin shows the catalogue and can grant the entitlement. The behaviours the catalogue describes, such as an AI voice receptionist or spam call detection, have not been confirmed as running, so treat this as a roadmap item and ask us before you plan around it.

  • WhatsApp AI Agents, Marketing Automation, Sales CRM, AI Call Receptionist
  • Catalogued with channels and default limits; the admin can grant the entitlement
  • Runtime behaviour is not confirmed, so treat as roadmap

Partner Bridge & Migration

bridge

A signed, retried event feed to partner products, plus importers to move a catalog, customers and orders onto QuantumOS.

Bridge sends events from your QuantumOS business to a partner product as signed, versioned messages that are queued safely and retried if the partner is down. Migration is a separate set of importers: an OpenCart importer for categories, products, customers and orders, a Shopify product importer, and catalog scrapers for Shopify, WooCommerce, OpenCart and Wix sites. These are one-time imports, not a live two-way sync with your old platform.

Transactional outbox

An event is saved in the same database transaction as the change that caused it, so it cannot be lost if the app crashes.

When something changes that a partner should hear about, the event is written to an outbox table inside the same transaction as the change itself. A drain job then sends the queued events to the partner. If the transaction fails, no event is queued, and if the partner is down, the event waits.

  • Event written in the same transaction as the change
  • Nothing is queued if the change is rolled back
  • A drain job sends queued events to the partner

Signed, versioned envelopes

Every message carries an id, a type, a payload version, a revision and an HMAC signature.

Messages use a neutral envelope with a unique event id the receiver can use to ignore duplicates, a payload version, the originating system, the time of the change and a revision number used to resolve conflicts in favour of the latest write. The signature includes a timestamp and the receiver rejects messages outside a five-minute tolerance. Payloads are validated against a schema for each event type.

  • Event id for duplicate detection
  • Payload version and per-type schema validation
  • Revision number; the latest write wins
  • HMAC signature with a five-minute timestamp tolerance

Partner and tenant mapping

Each message names the business on both sides, and unmapped pairs are rejected.

Every envelope carries the QuantumOS business id and the partner's business id. The receiver routes through a mapping table and rejects any pair that is not mapped, which stops a message from reaching the wrong business. Today the bridge connects QuantumOS to 1mp.space, with a related publishing path for Twetra.

  • Both business ids on every message
  • Unmapped pairs are rejected
  • Connects QuantumOS to 1mp.space, with a Twetra publishing path

Retries and the drain job

Failed deliveries are retried with a growing delay, and messages that keep failing are marked dead.

The drain job runs every minute, takes a batch of queued events and sends each with a 12-second timeout. A failed event is retried up to five times with a delay that starts at 30 seconds, doubles each time and is capped at one hour. After five attempts it is marked dead for a person to look at. Retries do not add random jitter, and delivered rows are marked as sent rather than deleted.

  • Drain job every minute, in batches
  • Five attempts, with delays from 30 seconds up to one hour
  • Events that keep failing are marked dead
  • No per-partner dashboard or alerts yet; the job logs its counts

OpenCart importer

Migration

Import categories, products, customers and orders from an OpenCart store.

The OpenCart importer reads from a MySQL database, a database dump or a hosting-panel backup, and brings categories, products, customers and orders across. The Value Store moved off OpenCart onto QuantumOS. An import is a one-time copy, and your team should check the result against the source before switching over.

  • Sources: live MySQL, a dump file or a hosting-panel backup
  • Imports categories, products, customers and orders
  • One-time import; verify counts and samples before cutover

Shopify and Wix imports

Migration

Bring products over from Shopify through the API or a CSV export, and from Wix.

The migration studio reads Shopify through its API or a CSV export, and Wix through its API, and imports the catalog. A CSV source is also available for any other platform. Wix catalogs have been imported for VisualPoint. These importers move catalog data; they do not keep a live link back to the old store.

  • Shopify through the API or a CSV export
  • Wix through its API
  • A generic CSV source for other platforms
  • Catalog import, with no live link to the old store

Catalog scrapers

Migration

Read a public Shopify, WooCommerce, OpenCart or Wix storefront to bootstrap a catalog.

Scrapers can read the public pages of a Shopify, WooCommerce, OpenCart or Wix store to create a starting catalog when you do not have database access. Scraped data is only as complete as the public pages are, so it is best treated as a draft to review.

  • Shopify, WooCommerce, OpenCart and Wix storefronts
  • Works from public pages, with no database access needed
  • Treat the result as a draft to review

Connector library

Adapters for payments, carriers, messaging, ads and accounting that your business can connect.

The connector library holds adapters for Razorpay, Stripe, PayPal, Cashfree and PhonePe; carriers including Delhivery, Blue Dart, DTDC, Shiprocket, Shadowfax, FedEx, DHL, UPS and USPS; WhatsApp providers WATI, Gupshup, AiSensy and MSG91; Telegram and Slack; Google Ads, Google Merchant Center, GA4 and Meta Ads and Conversions API; and Tally and Zoho Books. An adapter existing does not mean every operation is built or switched on for your business: some are limited to refunds and lookups, for example, so ask us about the specific flow you need.

  • Payments: Razorpay, Stripe, PayPal, Cashfree, PhonePe
  • Carriers: Delhivery, Blue Dart, DTDC, Shiprocket, Shadowfax, FedEx, DHL, UPS, USPS
  • Messaging: WATI, Gupshup, AiSensy, MSG91, Telegram, Slack
  • Marketing and analytics: Google Ads, Merchant Center, GA4, Meta Ads, Conversions API
  • Accounting: Tally and Zoho Books
  • Coverage differs by adapter; ask about the flow you need

EventOS

eventos

Sell tickets, take registrations and check people in for events and cinema screenings, in the same system as your store.

EventOS is for stores, venues and organisers that run events or screenings and do not want a separate ticketing platform. Today it covers creating events, a public event list and event pages, ticket limits, registration, QR tickets and check-in. Cinema-style seat selection, with showtimes and seat holds, has been added on top. Your own business can run events from day one, and if you run a marketplace, so can every active seller, with no separate install. EventOS is a foundation rather than a finished event suite, and the build status below lists what is not built yet.

Five plans, from free RSVP to multi-organiser marketplaces

Foundation

Lite, Pro, Hybrid, Enterprise and Marketplace plans match the kind of event you run.

Lite is for free RSVP events and check-in. Pro adds paid ticketing. Hybrid adds online and offline access with sessions. Enterprise adds speakers, sponsors, an AI assistant and API access, and Marketplace adds events from several organisers. The plans describe what each tier is meant to include. Some of those features, such as sponsor management, the AI assistant and public API access, are not built yet and are listed under build status.

  • Lite: free events, workshops, meetups and community sessions
  • Pro: paid workshops, training, shows and seminars
  • Hybrid: online, offline and hybrid events
  • Enterprise: expos, festivals, universities and summits
  • Marketplace: event platforms with several organisers

Capacity, pricing and QR tickets

The system decides whether a ticket request fits the remaining capacity, prices a registration with discounts and tax, and issues a signed QR ticket.

Before tickets are held, the system rejects invalid quantities and attempts to oversell, and it says exactly why. A registration is priced across several ticket types, with a limit on the discount and tax added on top. Each ticket carries a signed QR code that identifies the event, the booking and the attendee, and the signing secret is never inside the code itself.

  • Oversell is rejected before anything is held
  • Carts with several ticket types, a limited discount and tax
  • Signed QR tickets
  • Check-in is recorded for audit

Run events yourself, or for your sellers

Your own business can run events from day one. If you run a marketplace, every active seller can run events too.

An event's organiser can be your own business or a marketplace seller. A store with no marketplace sellers is never pushed into setting up marketplace features, and marketplace sellers get EventOS without a separate install.

  • Works for stores with no marketplace
  • Marketplace sellers can run events with no separate install
  • Back-office screens for events and for marketplace sellers
  • Seller hub access, plus the public events list

Cinema-style seat selection

New

Movies, venues, screens, showtimes and seat holds, with a seat picker and checkout for the public.

Staff add movies, create screens with a seat layout and schedule showtimes, and see how many seats are booked or held. Customers pick seats on your storefront, hold them while they pay, and receive a QR ticket. Checkout accepts either a showtime with chosen seats or a plain event with a ticket type, and it is protected by a bot check.

  • Movies, venues, screens with seat layouts, and showtimes
  • Live booked and held seat counts for staff
  • A seat picker for customers, with seats held during checkout
  • QR ticket page for the customer

Back office and public pages

Staff manage events, check-in and cinema showtimes in the back office. Customers browse events and open their tickets on your storefront.

Admins create events and run check-in from one screen, and manage movies, screens and showtimes from another. On the storefront, customers see a list of published events, an event page for each, and a ticket page that shows their issued ticket.

  • Back office: events and check-in, plus a separate cinema screen
  • Storefront: events list, event page and ticket page
  • Registration and checkout on the storefront

HotelOS

hotelos

Run rooms, rates and reservations for one or more properties in the same system as the rest of your business, and take bookings on your own website through a booking API.

HotelOS is for hotels that want their rooms, rates and reservations next to the rest of their commerce, not in a separate tool. Staff set up properties, room types and individual rooms, price each rate plan by date, and see availability on a calendar. Each stay becomes a reservation with its rooms, taxes and billing folio. A booking API lets your own website check availability, hold a room, take a verified payment, and look up or cancel a reservation by confirmation code. It does not include a ready-made public booking page.

Properties, room types and rooms

Foundation

Set up each property with its room types and the individual rooms behind them.

Inventory is held at two levels: room types, such as a deluxe double, and the physical rooms of each type. That lets you sell by room type and still assign a specific room to a stay. You can run more than one property in the same system.

  • More than one property in one system
  • Room types and physical rooms both tracked
  • Tax rules set up once and applied as separate lines on each stay
  • Setup screens for properties, room types and rooms

Rates by date

Price each rate plan for each date, so seasonal and promotional pricing is a price change, not a development job.

A rate plan describes what is being sold: whether it is refundable, its cancellation policy and what is included. A rate calendar holds the price for each date the plan is sold. Updating a block of dates does not change the plan itself.

  • A price for each date, not one flat nightly rate
  • Refundable and non-refundable plans, each with its own inclusions
  • A cancellation policy is defined once and reused across plans
  • A rates screen in the back office

Holds that keep the quoted rate

When a guest starts a booking, the room is held at the exact rate plan and party size they were quoted.

A hold records the rate plan, the number of adults and children, and the children's ages. Checkout then confirms that same rate and occupancy, instead of guessing or trusting figures sent with the request. Holds are short-lived by design.

  • The hold records the rate plan and the party size
  • Checkout cannot confirm a different rate or occupancy than the one quoted
  • Holds expire on their own

Availability and reservations

Staff manage availability, reservations and settings from one back office.

The back office has a dashboard, an availability calendar, a rates screen and a reservation list with a detail view for each booking. A reservation carries its rooms, its tax lines and its billing folio, and every change of status is recorded, so there is a trail of what changed and when.

  • Dashboard and availability calendar
  • Reservation list, and a detail view for each booking
  • Status moves through pending, confirmed, checked in, checked out, cancelled or no-show
  • Billing folio and status history kept for each stay

Booking API for your own website

Your site can check availability, hold a room, take a verified payment, and look up or cancel a reservation by confirmation code.

These functions are offered as a booking API that your developers connect to your site. Pricing and tax come from the same calculation in the API and in the back office, so a stay costs the same wherever it is priced. Today this is an API only: there is no ready-made public booking page.

  • Check availability and list properties
  • Hold a room, then complete checkout with a verified payment
  • Look up or cancel a reservation by confirmation code
  • No ready-made booking page ships today

B2B Commerce

b2b

Give trade customers their own portal to request quotes, reorder, route purchases for approval and track invoices, while your team runs pricing, credit and orders from one back office.

B2B Commerce is for businesses that sell to other businesses on agreed prices, approvals and credit terms, not just a card at checkout. You approve each customer, and they sign in to a portal that shows their own prices. They ask for quotes, send purchase orders, reorder and follow shipments on their own. Your sales and finance teams review quotes, set prices, credit limits and payment terms, and manage receivables from the same system that holds the orders.

A portal for each customer

Foundation

Approved trade customers get their own sign-in, with their prices, quotes, orders, invoices and documents in one place.

You approve each business before it can see any prices or orders. Once approved, its buyers see a dashboard of open quotes, pending orders and available credit, along with the business's addresses and tax details.

  • You approve each business before it sees commercial data
  • Dashboard of open quotes, pending orders and available credit
  • Billing, shipping and branch addresses saved on the account
  • GSTIN and PAN kept on the business profile

Quotes

Buyers ask for a quote, your team reviews and issues it, and the buyer accepts or rejects it online.

A quote request lists the products, quantities and requirements. Your team reviews price, stock and lead time before the quote goes out. Buyers follow each quote from draft to accepted, rejected or expired, and accepting a quote starts the order.

  • Request a quote for several products at once
  • Price, stock and lead time are reviewed before a quote is issued
  • Revisions are recorded without losing the quote's history
  • Expired quotes can be renewed for a fresh review

Prices that match each account

Show each customer the prices you agreed with them, including volume tiers and validity dates.

Set prices for a single account, with start and end dates, and offer lower unit prices as quantities go up. If a price moves after a buyer's approval, the buyer can ask you to review it.

  • Account-specific prices with start and end dates
  • Unit prices that step down by quantity
  • Buyers can ask for a price review after approval
  • Watchlists flag the products that dropped in price

Approvals for team purchasing

Staff at the customer raise purchase requests, and an approver signs off before an order is placed.

Several people at one customer can share an account with their own roles. A requester raises a request and tags it to a cost centre, and it goes to an approver. Order-value limits send larger purchases to the right level, and an approver who is away can hand over their authority for a set period.

  • Owner, requester, approver and finance roles
  • Approval limits based on order value
  • Cost centre on every request
  • Temporary delegation when an approver is away
  • A record of who approved each purchase

Purchase orders, reorders and recurring orders

Buyers send purchase orders, reorder past purchases in a few clicks, and set up recurring orders for items they buy often.

Purchase orders are checked against the account's rules before they reach your team, and the buyer's PO reference and document stay with the request. Buyers can ask to change or cancel an order until it is locked for fulfilment, and they can pause or resume a recurring order at any time.

  • Purchase orders checked against the account's rules
  • Reorder from order history in a few clicks
  • Recurring orders that the buyer can pause and resume
  • Change and cancellation requests before fulfilment
  • Export orders to a CSV file

Credit, payment terms and invoices

Sell on account with approved payment terms and credit limits, and give buyers clear invoices and statements.

You assign NET payment terms and a credit limit to each account, and orders that would go over the limit can be flagged or blocked. Buyers see each invoice as paid or outstanding, download GST tax invoices and view one statement for the whole account. They can ask for a higher credit limit and tell you when they have made a payment, so your finance team can check it against the bank.

  • NET payment terms and a credit limit on every account
  • Orders over the limit can be flagged or blocked
  • Paid and outstanding invoices, with GST tax invoices to download
  • Each invoice keeps the terms that applied when it was issued
  • Consolidated account statement
  • Buyers can request a credit review

Delivery and after-sales service

Buyers track shipments and raise service requests from the portal instead of phoning or emailing.

Buyers follow each order, including orders that ship in several parts, with serial numbers and warranty details kept alongside it. When something needs doing, they can request a delivery time window, installation help, a return authorisation or an escalation. Your team handles these requests in the back office.

  • Shipment status per order, including split shipments
  • Serial numbers and warranty details kept with the order
  • Request a delivery window, installation, a return or an escalation
  • Delivery preferences saved on the account

GST, terms and paperwork

Keep tax documents, agreed terms and vendor paperwork where a customer's procurement team can find them.

Quotes and invoices carry GST details, and a buyer can submit a tax exemption certificate for approval. Buyers accept your current commercial terms online, and each acceptance is recorded. Your vendor onboarding pack is in the portal, and buyers can download their quotes, purchase orders and invoices together.

  • GST-ready quotes and invoices
  • Tax exemption certificates submitted for approval
  • Acceptance of your terms recorded online
  • Vendor onboarding pack available in the portal
  • Download quotes, purchase orders and invoices together

Spend and savings reports

Show each customer what they spent and what your negotiated prices saved them.

The buyer's analytics page covers the last 12 months: total spend and order count, savings against your reference prices, how many quotes they accepted, and their current credit balance, with a chart of spend by month.

  • 12 months of spend and order count
  • Savings from negotiated prices
  • Quote acceptance rate
  • Current credit balance

Each business's data stays separate

Every B2B record belongs to one business, and every request is limited to the business making it, so one business does not see another's buyers, prices or orders.

The separation covers buyers, prices, quotes, orders and invoices. B2B Commerce runs today on the shared QuantumOS platform. Dedicated-infrastructure deployments are not supported yet.

  • Buyers, prices, quotes, orders and invoices kept apart per business
  • Every request is limited to your business
  • Runs on the shared platform; dedicated deployments not supported yet
Also shipping

Earlier-stage surfaces

Not everything is a fully documented pillar yet. This module is live in the admin today, with a smaller surface than the pillars above, and we are labelling it accordingly rather than giving it a launch page it has not earned.

Beta

SchoolOS

Class rosters, attendance, fee invoices, and exam results are stored by the school module. The admin UI covers a dashboard and a student roster today; timetabling, report cards, and parent-facing views are not built yet.

QuantumOS Dispatch

Insights for commerce operators

Operator stories, platform deep dives and notes on what ships.

We'll send the QuantumOS Dispatch by email. Unsubscribe any time. See our privacy notice.