IDA is an open source platform that gives people, organizations, and AI agents a verified, cryptographic identity they own — with no middlemen, no vendor lock-in, and full transparency.
"AI is acting in the world on your behalf.
Right now, no one can prove it was authorized to."
The internet gave us the power to connect.
It forgot to give us the power to prove who we are — or who's acting for us.
IDA is built on W3C Decentralized Identifiers — a standard that lets anyone create a cryptographic identity anchored on a blockchain, with no company in the middle. Think of it as a passport that no government can revoke, no platform can delete, and no AI agent can forge.
A Decentralized Identifier (DID) is a small string like did:adi:0x… that you own through a private key. No email address. No password reset email. Just math. Register once on the ADI blockchain; it's yours forever.
A university, employer, or government body issues a tamper-proof digital credential to your DID. You hold it in your wallet — like a physical card, but one that can be verified instantly anywhere in the world, without calling anyone.
Verifiable CredentialsWant to prove you're over 18 without showing your birthdate? Prove you're a licensed doctor without revealing your hospital? Zero-knowledge proofs let you share exactly what's needed — nothing more.
Zero-knowledge proofs
Every AI agent deployed through IDA gets a did:adi:agent:… — its own cryptographic identity, linked back to the human or organization who authorized it, with a complete audit trail.
When you authorize an AI agent to act for you, that authorization is a cryptographically signed token with exact scope — "buy groceries, max $200/week." The agent can't exceed that. And any service it talks to can verify the chain: you → delegation → agent.
Capability tokensAn Agent Trust Registry tracks every agent's history — verified actions, compliance attestations, autonomy level. This isn't a star rating you can game. It's a public, tamper-proof record. Rogue agents can be spotted. Well-behaved ones earn greater autonomy.
Trust registryThree forces are converging in 2025–2026 that make verifiable identity not a nice-to-have, but an operational necessity for any serious AI deployment.
High-risk AI systems must maintain traceable, auditable identity logs. If your AI agent made a decision that affected someone's loan, healthcare, or employment — you need to prove it was authorized, what it was told to do, and who was responsible. IDA is the identity layer that makes this technically possible.
By 2026, every EU citizen will have access to a digital identity wallet. IDA's architecture is fully aligned with this model — and extends it to AI agents. Governments, banks, and healthcare providers need open infrastructure that speaks the same language.
Anthropic's Model Context Protocol (MCP) and Google's Agent-to-Agent (A2A) protocol are rapidly becoming how AI systems talk to each other. IDA integrates with both — adding verified identity to every agent-to-tool and agent-to-agent interaction.
AI agents are being deployed to book travel, execute trades, manage files, and handle customer service — all without a human in the loop for each action. The identity and accountability gap is not theoretical. It's happening today, and the consequences of getting it wrong are real.
Universities are issuing digital diplomas. Healthcare systems are exploring portable patient records. Professional bodies want machine-verifiable licenses. The missing piece is a universal, open standard for issuing and verifying these credentials — without locking into any single vendor.
When identity is controlled by a handful of tech platforms, those platforms become chokepoints for speech, commerce, and civic participation. Open, decentralized identity infrastructure is a public good — like roads, not like toll booths.
If you're shipping agents that act on behalf of users — shopping, research, automation — you need a way to prove authorization without asking users to share API keys. IDA gives your agents a verified identity out of the box, with SDKs in TypeScript, REST APIs, and full Docker/K8s deployment guides.
Need EU AI Act compliance? Deploying a multi-agent workflow across departments? IDA's delegation chains and audit logs give your legal and compliance teams what they need: a cryptographic proof of authorization at every step, with no black boxes.
Issue tamper-proof digital credentials that your recipients own forever, can share selectively, and can verify anywhere — without calling your office. IDA's issuer portal handles bulk issuance, revocation, and analytics on how credentials are being used.
IDA implements DID Core, VC Data Model v2.0, DIDComm v2, and zero-knowledge proof circuits (Groth16, Semaphore). The codebase is fully open, the smart contracts are auditable, and we're actively looking for contributions to the did:adi method spec.
Identity infrastructure is too important to be owned by a single company. IDA is licensed under Apache 2.0. The smart contracts are public. The DID method spec is open for comment. We believe the world's identity layer should be a commons — built in the open, auditable by anyone, and improvable by everyone.