IDENTITY & ACCESS FOR AI AGENTS

Backed by

Combinator

Every agent needs an identity.
And clear permissions.

Give every AI agent an identity. Control which tools and data it can access, what it can do, and see every action it takes.

See Fabriq in action →

BUILT FOR AGENTS. GROUNDED IN TRUST.

IDENTITY / PERMISSIONS / AUDITMANY AGENTS. ONE FABRIC.MOVE TO EXPLORE · DRAG TO ROTATE

EVERY CONNECTION, ACCOUNTED FOR.

IdentityPermissionsAudit

01 / WHERE WILL YOUR AGENT RUN?

One fabric. Two ways in.

01 / FOR YOUR ORGANIZATION

Inside your company.

Agents that act for your employees, with each person’s permissions and every call accounted for.

Classic painting of a sunlit porch swing — 01 / A familiar place to work

Book a demo Explore internal agents

02 / FOR YOUR CUSTOMERS

Inside your product.

Your customers connect their own accounts. Fabriq handles the credentials, so your product never holds a token.

Classic landscape painting of open fields under cloud — 02 / Room to explore

Start now Explore the developer platform

THE CHALLENGE

Your agents can do more.
Your controls should, too.

Agents cross tools, act for people, and make decisions. Traditional identity systems were built for a different kind of work.

One shared key. Too much access.

Policies in prompts. No enforcement.

Actions everywhere. No shared record.

THE REQUEST PATH

One connection.
Control at every boundary.

Instead of agents connecting directly to every system, they route through Fabriq. Authentication, credentials, and governance travel with each request.

FOLLOW A REQUEST

YOUR AGENTS

sales-agent for Dana · Sales

support-bot for Priya · Support

data-pipeline for Alex · Admin

Agentic Fabriq

Identity · credentials · policy

Identity verified · Policy allowed

01 Identity02 Policy03 Audit

TOOLS & SERVICES

Gmail ✓ Allowed

GitHub × Restricted

Database ✓ Allowed

✓ ALLOWEDsales-agent · dana@acme · gmail.send

Identity verified. Both agent and user have gmail.send. Request forwarded and logged.

Streamline auth

One connection to your agent tools and data. Tokens stay in the vault, outside agent context.

Control access

Resource-level scopes per agent and per user. Permissions are enforced at the tool boundary.

Follow every action

See the acting identity, policy decision, and request metadata in a single audit trail.

02 / THE CONTROL LAYER

Every action checked
against policy.

Which agent, for which person, against which resource. Decided before anything runs.

Swipe to explore all six tools →

AGENT × TOOL	Gmail	Slack	Drive	GitHub	Database	Payments
sales-agentfor dana@acme
support-botfor dana@acme
data-pipelinefor dana@acme
finance-agentfor dana@acme
research-agentfor dana@acme
ops-agentfor dana@acme

✓ Allowed

sales-agent → Gmail, acting for Dana. Both the agent grant and the user’s permission allow this tool.

● ALLOW × BLOCKEDIllustrative policies · effective access = agent grants ∩ user permissions

SET UP IN AN AFTERNOON

Three commands,
or one pasted prompt.

Register the app your agents run under, let each user approve their own tools on a page we host, then mint a short-lived token and call through the gateway.

01
Register your agents
An app id, a display name and the scopes it may ever ask for.
afctl applications register --app-id my-agent
02
Connect each customer
They approve the provider’s own consent screen. No Fabriq account, and you keep your own user ids.
POST /oauth/{provider}/initiate
03
Mint a token, call tools
A token scoped to one of your users, then tool calls through one endpoint.
await client.list_tools()

03 / CONNECT ONCE

The tools your agents
already use.

Explore all integrations

04 / WHAT WE DO

Everything your agents
need to run safely.

One platform for secure, governed, and observable agent deployments.

1connection to your agent tools

100%of actions logged & audited

0secrets stored by agents

01 / Identity

Agent & user identity management

Create a unique identity for every agent. Carry the acting user’s identity through every request, so delegated work stays within that person’s permissions.

Agent creation and lifecycle management
Delegated user identity with permission constraints
Identity propagation across tools and services

Explore identity

02 / Access control

Fine-grained authorization & policies

Define precisely what each agent can do, which resources it may touch, and whose authority it acts under. Enforce those rules at the tool boundary.

Resource-level access policies
Consistent governance across workflows
One connection to your tools and data

Explore access control

03 / Security

Credential & token management

Keep sensitive credentials out of agent context. Connect your existing identity provider and let Fabriq handle authentication, token exchange, and secure storage.

Secure credential vault and token exchange
Agent and user authentication verification
Drop-in integration with existing infrastructure

Explore security

04 / Visibility

Full audit logging & monitoring

See the tools, data, and users behind every action. Follow a request from its identity and policy decision to its outcome.

Real-time activity visibility
Immutable audit logs for compliance
Incident investigation and request tracing

Explore visibility

05 / Developer experience

Simple integration. Instant control.

Add identity and governance to the agents you already build. Use SDKs and APIs without redesigning your infrastructure.

SDKs for major agent frameworks
REST APIs for custom workflows
One governed MCP connection

Explore developer experience

05 / HOW IT WORKS

From zero to secure
agents in minutes.

LangChainCrewAIAutoGenOpenAI Agents

FABRIQ CONSOLEInteractive preview

sales-agent-v2 / STEP 01

Register your agents

Every agent has its own identity. Select an agent, or add one to this preview.

Step 1 of 5

06 / BUILT FOR THE NEXT SHIFT

Every computing shift
reshaped identity.

2000s

SaaS

Single sign-on

2010s

Cloud

Identity & access management

2015s

APIs

Delegated access with OAuth

Now

Agents

Fabriq

FROM THE JOURNAL

Ideas for an agentic world.

Classic painting used as the article cover

AGENT OPERATIONS

Building Trust in Autonomous Systems

Trust doesn't come from claiming an agent is safe. It comes from proving it's governed — scoped, watched, recoverable, and owned by someone who can answer for it.

Read the story

DELEGATION

Agent-to-Agent Authorization: What Breaks When Delegation Chains Get Long

A2A standardizes how agents talk, not what authority travels with the request. What breaks at hop three, which mechanisms exist today, and the invariants worth enforcing before they do.

Read the story

AI Safety

Architecting Against Prompt Injection: Why Filtering Fails and Structure Works

Detection-based defenses plateau under adaptive attack. The durable answer separates control flow from data flow so untrusted content structurally cannot choose which privileged action runs.

Read the story

LET’S BUILD YOUR BOUNDARIES

See what your
agents could do.

30 minutes with our founding team. Your workflows, your questions, and a closer look at Fabriq.

See the full product in a live demo
Get answers from our founding team
Free onboarding assistance
No commitment required

BACKED BY

BUILT BY

Every agent needs an identity.And clear permissions.

One fabric. Two ways in.

Inside your company.

Inside your product.

Your agents can do more.Your controls should, too.

One shared key. Too much access.

Policies in prompts. No enforcement.

Actions everywhere. No shared record.

One connection. Control at every boundary.

Agentic Fabriq

Streamline auth

Control access

Follow every action

Every action checked against policy.

Three commands, or one pasted prompt.

The tools your agentsalready use.

Everything your agents need to run safely.

Agent & user identity management

Fine-grained authorization & policies

Credential & token management

Full audit logging & monitoring

Simple integration. Instant control.

From zero to secure agents in minutes.

Register your agents

Every computing shiftreshaped identity.

SaaS

Cloud

APIs

Agents

Ideas for an agentic world.

Building Trust in Autonomous Systems

Agent-to-Agent Authorization: What Breaks When Delegation Chains Get Long

Architecting Against Prompt Injection: Why Filtering Fails and Structure Works

See what youragents could do.

Every agent needs an identity.
And clear permissions.

Your agents can do more.
Your controls should, too.

One connection.
Control at every boundary.

Every action checked
against policy.

Three commands,
or one pasted prompt.

The tools your agents
already use.

Everything your agents
need to run safely.

From zero to secure
agents in minutes.

Every computing shift
reshaped identity.

See what your
agents could do.