SKU updates per second across the network
p99 latency from source change to subscriber
Channels, AI agents, marketplaces and feeds supported
From paste-in to live feed for a new retailer
The Challenge
Stale data breaks AI
In the AI age, every shopping question is answered with confidence. If the data behind that confidence is a day old, every booking, click and recommendation is wrong before it lands.
Confidence problem
LLMs and shopping agents do not flinch. They name a product, quote a price, recommend an alternative. Last week's feed becomes confidently wrong.
Conversion problem
A click that lands on an out-of-stock product is a lost sale, a cold lead, and a dent in the trust the customer extends to every assistant they will ever use.
Syndication problem
A nightly CSV for Google Shopping, a spreadsheet for an affiliate, a different mapping for every marketplace. The shadow pipeline becomes the weakest link.
Verification problem
A good agent should be able to cite its source, check a freshness stamp, and refuse to answer if the data is older than it trusts. None of the existing pipes give it that.
The Opportunity
A new layer of the internet, just appearing
Every retailer already runs a tangled mess of feeds. AI assistants and shopping agents are now joining the queue, demanding live, verifiable data with structured freshness guarantees. There is no incumbent layer for that. There needs to be.
Global retail, the catalogue most agents are about to reason on top of
Major AI shopping surfaces (ChatGPT, Perplexity, Claude, Gemini, Rufus) launched in 18 months
Existing data layers built for second-level freshness across all of them at once
Merchand sits between a retailer's catalogue and every downstream surface that needs to know the truth. One source, signed and timestamped. Many destinations, each in the shape they expect. Updates fan out in seconds, not overnight.
Our Approach
What we delivered
A complete live data platform: ingestion, recipe-based normalisation, fan-out to every downstream surface, plus an AI layer that turns raw catalogue rows into channel-ready, search-tuned, on-brand product content.
Live ingestion engine
Connectors for Shopify, WooCommerce, BigCommerce, custom APIs, XML, CSV, Google Sheets. AI ingestion wizard auto-detects schemas, maps fields, and writes deltas to an event bus the moment the source changes.
Visual recipe editor
Drag-and-drop transforms turn raw rows into clean product records: concat, derive, replace, regex, conditional logic, currency normalisation, dedupe keys. Versioned per retailer, previewable on real rows.
Multi-format fan-out
Every subscriber pulls the fields it expects in the format it expects: CSV feed, signed webhook on every delta, or live LLM-ready API with monotonic as_of timestamps. One source, many projections.
AI feed optimisation
Titles rewritten for search intent. Descriptions generated on brand voice. Categories auto-mapped to Google taxonomy. Attributes extracted from copy. Feed health audited and explained in plain English.
Closed-loop freshness
A signed JS tag on retailer product pages reports back the price and stock customers actually see. Any drift triggers an immediate re-import that fans out to every subscriber within seconds.
Marketing and brand layer
A/B title experiments per channel with auto-promotion, on-brand AI imagery and 8-second video ads at catalogue scale, per-publisher governance, click analytics, and a full audit log of every delivery.
The Results
A foundational layer of modern retail
Merchand launched as the live data infrastructure for retailers, publishers and AI agents. One catalogue, every destination, second-level freshness, end-to-end signed.
p99 fan-out latency
From a price change at the source to every downstream subscriber, signed and verifiable, in 38 milliseconds at the 99th percentile.
Live destinations
Google, Meta, TikTok, Amazon, Pinterest, Shopify, ChatGPT Shopping, Perplexity, Claude, Rufus, plus affiliate networks and custom APIs.
From paste to live
A retailer hands us a URL on Day 0, and on Day 5 the feed is clean, on every channel, governed and observable. No engineering team required.
“Every retailer already runs a shadow pipeline of nightly CSVs and one-off integrations. AI agents are about to make that the most expensive infrastructure mistake of the decade. Merchand is the layer we wish had existed when we first started building feeds twenty years ago.”
Keith Frangleton, CEO
Scope
Services delivered.
A full-stack platform build spanning data infrastructure, AI engineering, marketing site design and product UX.
Live Data Infrastructure
Event-driven ingestion, change detection, recipe-based normalisation and signed fan-out
AI Product Engineering
AI ingestion wizard, feed optimisation, A/B title experiments and on-brand creative generation
Channel Integrations
Native templates for 142+ channels, AI agents, marketplaces, affiliate networks and custom APIs
Platform Engineering
Multi-tenant SaaS architecture, signed events, governance, audit logging and observability
Brand and Marketing Site
Identity, wordmark, palette system, illustration direction, marketing copy and the merchand.io site itself
Product UX and Onboarding
Recipe editor, ingestion wizard, publisher dashboards, retailer setup flow and end-to-end product design
