Skip to content
Blog
Company

Introducing AmanTrue: the control layer for institutions deploying AI

Why we exist, and why every AI agent needs a passport.

A robotic hand reaching for a stack of records, stopped by a barrier in between

Who we are

Today we are introducing AmanTrue: the control layer for institutions deploying AI. We are built in Abu Dhabi, at Hub71, by the founders and team behind identity software already in production at banks and governments.

Institutions are putting AI agents to work on their most sensitive records. AmanTrue makes sure those agents stay under their control. AmanTrue Agent Passport proves who an agent is. AmanTrue BorderControl decides what it can see and do.

In this post we explain why AmanTrue exists, what our two products do, and who is behind them.

Why we exist

Imagine an AI agent just leaked a record from your system. Whose agent was it? In most institutions deploying AI today, nobody can say.

For decades, a clerk and an application stood between people and an institution’s records. The clerk was a known employee. The application knew who was logged in, what they were allowed to see, and kept a record of who did what. AI agents now take over that work. They talk to citizens and customers, read records and complete tasks, around the clock.

But an agent does not carry a name the way a clerk did. Picture this: your AI agent is logged in as your finance manager, or on a shared key the whole team uses. Nobody issued it an identity, and nobody vouches for it. Your logs show the key, not the agent, and not who sent it. The agent is a nobody almost everywhere.

Two ID badges side by side. The person’s badge has a photo and a name, Mohamed Ahmed. The agent’s badge is blank. Your agent has no ID.

The evidence shows how common this is:

  • Most organisations cannot name their own agents. In Okta’s 2026 survey of 306 security leaders, 47% were confident they could identify every AI agent in their organisation, and 46% could centrally control what those agents access.
  • Agents act beyond what anyone approved. In July 2025, an AI coding agent on Replit deleted a live production database during a code freeze, as reported by The Register.
  • You answer for what your agent says. In 2024, a Canadian tribunal held Air Canada liable after its support assistant gave a customer wrong information about a refund policy.
  • The UAE is moving fast, and control has not caught up. The head of cybersecurity for the UAE Government has said half of government services need to become agentic (Computer Weekly, GISEC 2026). In a Dataiku survey conducted by Harris Poll and reported by Gulf News, 80% of UAE CIOs said they had encountered an agent that violated business intent, and only 5% said they could reliably identify and contain a problematic agent within one to two hours.

When something goes wrong, the first questions are always the same: which agent did it, who sent it, was it allowed, and can you stop it? Today, most institutions cannot answer these questions. When business breaks, they ask you to explain, not the agent.

Every minute counts: only 5% of UAE CIOs can contain a problem agent within 2 hours. 80% have already seen an agent break business intent. Dataiku survey by Harris Poll, reported by Gulf News, 29 September 2026.

What we offer: Two products. One control layer.

AmanTrue is the control layer for companies, institutions and governments deploying AI. Before an agent touches a record, AmanTrue answers two questions: who the agent is, and what it can see and do. AmanTrue Agent Passport proves who the agent is. AmanTrue BorderControl decides what it can see and do.

AmanTrue Agent Passport: every agent, registered and accountable

Who is the agent, and who is accountable for it?

AmanTrue Agent Passport is the identity an authority issues to an AI agent. It ties the agent to the person or organisation responsible for it, records what that agent may do and until when, and can be revoked at any time. Every action leaves a proof an auditor can check.

Like a traveller at the border, every agent has to answer who it is, who sent it and what it may do:

  • Identity, like a passport: who the agent is, verifiable by anyone.
  • Delegation, like a visa: who stands behind it, what it may do, and until when.
  • Audit trail, like a border log: every action leaves a record.
An AmanTrue ID badge for Mohamed’s Agent 007, with a verified check mark. Now your agent has a verified ID.

AmanTrue BorderControl: what agents can see and do with your records

What can the agent see and do?

Once you know who the agent is, you decide what it may see. AmanTrue BorderControl sits between the agent and your records. The agent asks a question, such as whether a driving licence is valid, and gets the fact it needs, signed by the source. The record never leaves your custody. Your policy decides which facts an agent may ask for, and every question joins an audit trail.

BorderControl works with AmanTrue Agent Passport or with your own identity provider. Both answers signed. Both auditable, without opening the record. That is what lets your security team say yes.

Proudly built in Abu Dhabi and scaled to the world

The founders and team behind AmanTrue built identity and verification software that is already in production at banks, governments and global technology companies.

  • The founders. Evin McMullen founded Disco (acquired) and has been a director at ConsenSys and Berkshire Hathaway. David Suarez built Hermez, acquired for $250M, and was CTO of Polygon.
  • Where the team’s technology already runs. Banks, the Government of India (delivered with Chaincode), a European government under NDA, and the EU regulatory sandbox on data verification.
  • At scale. 5M+ people use identity systems this team built. 9,000+ companies and projects build on its verification software. 60+ organisations served, including governments, banks, Big Four firms and telecoms. Microsoft, TikTok and Telefónica Tech run products on software this team wrote and still maintains.
  • Built here, for here, first. AmanTrue is based in Abu Dhabi, at Hub71, with 25+ engineers and senior technical leaders.

What is your biggest obstacle in deploying AI?

Tell us about your case at amantrue.com/contact. We read every message, and a 30-minute consultation with our team is the next step.

Sources

What is your biggest obstacle in deploying AI?

Tell us about your case. We’ll show you how it applies to your records, on your infrastructure.

Contact us