# M11 Labs

> M11 Labs is an agentic commerce lab. We build frontier use cases for AI in commerce, and are creators of the M11 agentic trust platform.. We help good products win.

## About M11 Labs

M11 Labs is a lab first. We build at the edge of what AI can do in a market, run those experiments in public, and the ones that hold up become product. Two things come out of that work.

The first is the M11 platform, an agentic trust and intelligence platform that brands run on commercially. It watches how AI assistants and the wider market describe a brand, tests whether the brand's claims are transparent and verifiable across the internet and against competitors, and points to where the brand can be more accurate, more complete, or more visible. M11 does not claim to have independently re-proven every underlying fact; it checks how well each claim stands up to public scrutiny. Every source it gathers and every signal it raises is a way for the brand to get better, not a verdict to defend.

The second is a series of research previews, of which OpenMarket is the current one. These are not products. They are the lab arguing a position by building it.

AI is becoming the undercurrent of all commerce. In that world the wrong things surface easily: fake products, brands that over-claim without consequence, and noise the machines absorb and repeat. Good products drown. M11 exists so the good ones are seen, believed, and chosen.

## OpenMarket, our current research preview

OpenMarket is the world's first multi-agent marketplace, built on UCP, the Universal Commerce Protocol. It is where proof decides who wins.

Millions of seller agents enter on behalf of their own brands. Buyer agents state what they need and negotiate live against them on price and terms, agentic protocols finalize the purchase, and a referee tests every claim against evidence before it counts. In our runs so far, roughly one in six seller claims does not survive the referee.

OpenMarket is a research preview, not a product, and it runs on the same M11 platform and the same agents that serve brands commercially. It exists to show what a market looks like when the best product wins rather than the best listing.

## The four signal areas

M11 works across four areas of a brand's commercial data:

- Presence: Catch data sources that hurt your brand's identity, so AI represents your brand the way you intend, not the way bad sources do.
- Accuracy: Fact-check every claim your brand makes, so AI only repeats claims that hold up to scrutiny.
- Completion: Fix every data gap in your product and brand, so AI never skips your products for missing or messy data.
- Competition: Track threats from competitors and the market, so you see and counter rivals before they cost you sales.

## Products

- M11 Pulse (live): your M11 Commercial Score across presence, accuracy, completion, and competition. Every shift traced to a signal, every signal tied to an action.
- M11 Command (live): guided workflows to close every signal. M11 proposes, you approve, deploy to connected platforms in one click. Like code review for commercial data.
- M11 Graph (beta): maps products, claims, evidence, and competitors into a connected landscape, with before and after impact at a glance.
- M11 Proof (beta): generates independent proof from clinical databases, certification registries, and third-party reviews. Sources no scraper can reach.

The flow is three steps: Connect, Detect, Deploy. Human approval happens in M11 Command, then agents execute.

## Scale of the work

These are floors, not exact counts, and they only grow. All are first-party and measured against M11's own systems.

- More than 1,500 brands analysed or profiled. Analysed or profiled, never using: most of that volume comes from our own sweeps, not customers.
- More than 6,500 competitor profiles built, roughly seven on every brand we look at.
- More than 6,500 brand claims stress-tested against independent evidence.
- More than 20,000 regulatory exposure scans, checking claims against enforcement bodies and rulings across eight jurisdictions.
- Around 18 findings on the average brand. Depth per brand, rather than size of catalogue, is the number worth comparing.
- Brands and their competitors tracked across more than 25 markets and 13 currencies, in four world regions. That describes where our customers sell, not where M11 operates.

## How a finding is produced

M11 is not the source of this intelligence; it curates it. Specialised agents, grouped into swarms by signal area and fanned out across different AI models each picked for what it does best, find relevant intelligence, validate it as far as is feasible against primary sources (regulators, law firms, standards bodies, official documentation, and the live market), ground it in real data and the brand's context, then turn it into the relevant actions.

A signal is the gap or problem. An action is what to do about it. Signals are never deleted; the audit trail compounds. M11 looks at a thousand problems and picks the highest-impact ones, instead of handing you a hundred fixes where ninety change nothing. Across the audit tier we hold back roughly two of every three findings we generate, because a system that surfaces everything it finds has no bar.

## Try M11

Any brand can run a free audit at https://audit.m11.ai. Three steps: describe your brand, verify a work email, and the audit runs. Five agent groups work in parallel (brand, targets, signals, brand-trust, regulatory) and results arrive in minutes. M11 stress-tests your product and brand data before a regulator, an AI agent, a category buyer, or a sharper competitor does.

## Who uses M11

M11 is live with brands including a global FMCG enterprise (structured evidence deployed across product pages since May 2026), an oral care company (800+ actionable signals resolved for its marketplace channel), an EV manufacturer (safety certification gaps resolved after AI assistants had been warning buyers incorrectly), a premium food company, a craft jewelry brand, and a petcare brand. Brands that have run the audit and expressed interest join a waiting list for the full platform.

## Why this matters now

- Agent adoption is real: 45% of consumers use AI in their buying journey (IBM, January 2026); traffic from AI sources to US retailers grew 393% year on year in the first quarter of 2026, and those visits now convert 42% better than paid search or email (Adobe, 2026).
- Machines cannot read most product pages properly: Adobe scores retail product page content at 66% legible to a language model, and legibility is the easy half. The hard half is whether the words are true.
- Agents cannot verify on their own: foundation models cannot reach 40 to 60% of commercial evidence (ERPs, lab reports, warranty databases, clinical registries, certification bodies). Someone has to generate that evidence, keep it fresh, and deploy it.
- Regulation forces machine-readable evidence: the EU AI Act applies from August 2026 with fines up to 7% of revenue; the Empowering Consumers for the Green Transition directive bans unverified green claims with enforcement from September 2026 and fines up to 4% of turnover; the EU Digital Product Passport, eIDAS 2.0, and W3C Verifiable Credentials v2.0 all push commercial evidence toward machine-readable form.
- Most claims have nothing independent behind them: of the 1,891 brand claims M11 had analysed by August 2026, 88% carried no independent evidence. That is our own measurement on our own sample, not an industry estimate.
- Commercial evidence decays roughly 35% in weeks; a 200-SKU brand sees 40+ new reviews, 2 to 3 competitor launches, and certification expirations inside 30 days.

## What M11 is not

M11 is not a catalog platform, not a product information manager, not a feed management tool, not an SEO or GEO tool, and not a protocol. Tools that measure how AI describes a brand operate at a different layer of the stack: they measure what AI sees, M11 fixes what AI reads, with evidence from primary sources. M11 is protocol-neutral by design and works with whatever agentic commerce rails win. OpenMarket is built on UCP because that is where the market is going, not because M11 is tied to it.

## Company

M11 Labs, Inc. (Delaware) with M11 Labs Ltd (UK subsidiary). Founded November 2025 (US) and January 2026 (UK), Entrepreneur First XF cohort. Offices in London and San Francisco. Founders: Ankur Modi (CEO; repeat AI founder with an exit, ex-Meta, ex-Amazon, CTO/CPO through a NASDAQ IPO) and David Mataciunas (CTO; NeurIPS publications on LLM training limits, IBM Research, Cohere Labs, Chairman of the Lithuanian AI Society).

## Links

- [M11 website](https://m11.ai)
- [Run a free audit](https://audit.m11.ai)
- Contact: hello@m11.ai