Agentic commerce is coming fast. trust.ai is the layer that makes every AI-made purchase, consent and contract provably backed by a real person, wherever the agent acts: inside an AI client, through a connector, or on a regular website. People stay in control. Merchants can finally rely on what an agent says.
01 The shift
People are starting to state outcomes instead of operating websites. 'Find me a cheaper energy plan and switch me.' The AI does the rest. The interface era is ending; the client era has begun.
AI browsers are here
ChatGPT Atlas, Perplexity Comet, Claude in Chrome, Copilot Mode: every major AI now ships or powers a browser.
1,400+
apps and connectors in ChatGPT's directory.
439
connectors in Claude's directory, up from a handful a year ago.
Checkout is live
Instant Checkout inside ChatGPT, powered by Stripe. Google and Shopify push UCP. The rails are being laid now.
Twelve months ago, none of this existed at scale.
02 The prize
Flights, energy contracts, insurance, software, ad spend, everyday shopping: everything bookable will be bookable by an agent. And merchants will chase those agents with AI-readable catalogues, agent-ready checkouts and ads placed inside AI clients. Marketing will target models as much as people.
USD 190 to 385bn
AI-agent driven share of US e-commerce by 2030 (Morgan Stanley)
1 in 2
online shoppers expected to buy through an AI agent by 2030, up from about 5% today
USD 3 to 5tn
projected global agentic commerce by 2030
03 The gap
Payments already have rails: tokens with limits that an agent can spend safely. But most of a transaction is not payment. It is accepting terms. Confirming that your name and address are correct. Consenting to data use. Proving your age. Signing a contract. Under EU law, and far beyond it, each of these is a legal act.
When an agent clicks 'I agree', who agreed? Today no merchant can prove a human stood behind that click. No proof means disputes, GDPR exposure and liability that lands nowhere.
The first chargeback is a payment problem. The first 'I never agreed to that' is a legal one.
04 The answer
trust.ai sits between people and every AI they use.
You decide, once, what your AIs may share: name, address, age, contact details, payment references. Verified once at bank level, kept under your control, revocable in one tap. Works across Claude, ChatGPT and AI browsers, and whatever comes next.
Every statement an agent makes arrives with a trust.ai token. One call verifies it: a real, identified person; accurate, verified data; a valid mandate covering exactly this action; consent recorded as evidence. You can rely on it, and you can prove it later.
Identified human
An eIDAS-grade identity check anchors every mandate.
Signed mandate
The user signs that statements made under the mandate are their own. That is the liability shift that makes reliance possible.
Evidence per event
Every consent and confirmation is logged and exportable, for the regulator as much as for the dispute.
In the AI client
Claude, ChatGPT and other assistants call trust.ai through a connector. Identity, data and consent arrive inside the chat.
On any website
No special integration needed. An agent working in an AI browser completes a purchase on a regular web shop and checkout, with trust.ai behind every step. Tested end to end.
Through APIs
Merchants that offer an agent interface verify the trust.ai token with one call, and get the evidence with it.
No pop-ups, no app switching. If you want, trust.ai brings terms, consents and the data being shared into your AI chat, in plain words. You choose how much you see: everything, nothing, or only what is legally binding. And you set rules your agent cannot override, for example 'never share my phone number' or 'always ask me before signing a contract'. trust.ai never accepts terms for you. You do, under the mandate you signed.
05 The magic
Agentic commerce gives every merchant two new funnels. trust.ai makes both of them convert, and both of them stick.
You reach the customer. An ad no longer links to a landing page. It hands the customer's AI a task.
Advertisement
Still overpaying for power?
GreenVolt Flex24. Save around 384 EUR a year.
Switch my electricity to GreenVolt Flex24. Use my trust.ai profile for identity, address and consents.
No ad needed. The user states an outcome, and the agent picks the merchant it can actually complete the deal with.
Agent-ready and trust.ai-verified means the sale is yours, not the competitor's whose checkout an agent cannot legally finish.
Copy or ask
A prompt from an ad, or a question in the client. Either way the agent gets a task.
Act
Your AI starts the switch and reaches the provider, through its agent interface or simply its regular website.
Trust
trust.ai supplies your verified data and records every consent, under the mandate you set up once. You see as much of it in the chat as you want. No app to open, no form to fill.
Done
A valid contract, evidence stored, and a provider that can rely on every statement. You never left the chat.
You authorise once. Only when a provider needs a data point you have never authorised, say the meter number for an energy switch, does trust.ai ask you to add it once. After that, everything that needs it runs entirely inside your AI.
No forms. No re-typing. No doubt about who agreed. Two funnels, one result: more completed sales for every merchant that participates.
06 Where we are
trust.ai is not a slide. It runs today, it has been tested end to end, and it is rolling out.
Live in the leading AI clients
The trust.ai connector runs in Claude and ChatGPT today: verified identity, signed mandate and an evidence log behind every single action.
Agentic purchase on any web shop
Tested end to end: an agent completes a purchase on a regular online shop under the user's mandate. No shop-specific connector, no integration project.
Signing from the chat
Qualified electronic signatures, completed inside an AI client with a passkey and an ID check. The strongest proof of a human there is, now available to agents.
Rolling out
The trust.ai mobile app is in beta testing, our listing in the AI app directories is in the approval process, and the first merchant agreements are being signed.
07 Why us
Trust infrastructure is not something you announce. It is something you are licensed, audited and battle-tested to run. Signicat has been doing exactly that for 20 years.
Qualified Trust Service Provider
Certified under the European eIDAS framework and listed on the EU trusted lists. Qualified signatures, seals and attestations are our regulated day job.
The eID backbone of Europe
One integration reaches 35+ national eID schemes. We run enrolment for national identities such as Norwegian BankID, and our identity checks are trusted in settings as critical as UK border processes.
20 years, zero shortcuts
Two decades of identity verification, KYC and signing for banks, insurers and governments. Compliance is not a checkbox for us; it is the product.
European by design
The agent platforms will come from everywhere. The trust layer for Europe has to take European law seriously: GDPR, eIDAS 2.0, consumer protection. That is precisely the role we are built for.
Making AI capable is the industry's race. Making it accountable is ours.
We are opening trust.ai to merchants, platforms and networks that want to sell safely in the agent channel. One partner of scale turns this from a product into the default.
Talk to us