Prebuilt connectors for the tools you run.
Microsoft 365, Google Workspace, Slack, and GitHub — connect once, grant per agent.
PLATFORM / Integration Hub
Connect your tools once. Every agent gets them.
One endpoint between your agents and your software. Connect an agent once, and every integration you’ve authorized comes with it.
01 / IDENTITY02 / POLICY03 / AUDIT
01 / THE CONTEXTConnect once.
Connect once.
Reach everything you are allowed to.
Every agent rebuilds the same plumbing — OAuth, tokens, API clients — for every tool it touches.
AUTHORITY
agent
∩
person
RESOLVED PER REQUEST∩
person
02 / IN PRACTICE
One connection. Many possibilities.
ONE CALL · ONE ENDPOINT · THREE TOOLSIllustrative · research-agent acting for Researcher
THE CALL
tools/call- agentresearch-agent
- acting forResearcher
- connectedonce, to Fabriq
One authorization behind it. The agent holds no token for any of the tools it reaches — the endpoint attaches those at call time.
Google DriveSearch approved sourcesdrive.search · the shared research folderAllowedInside the agent’s grant and inside the researcher’s own Drive access, so the call goes through with a vault-held token.
NotionExport restricted documentsnotion.export · Legal’s private spaceBlockedThe researcher cannot open that space, so neither can an agent acting for them. Stopped at the endpoint; Notion is never called.
SlackPublish a reportslack.post_message · #research-briefingsHeldA named human approver has to say yes first. Nothing is posted while the call waits.
Illustrative requests · every one recorded with both identities.
03 / WHAT CHANGES
Control, in the details.
Import an OpenAPI spec.
Your internal API becomes agent-usable tools automatically.
Plug in MCP servers.
Register a remote server and Fabriq discovers its tools.
Postgres with guardrails.
Read-only queries, checked before they ever run.
Reach behind your firewall.
An outbound-only encrypted tunnel — no inbound ports to open.
One endpoint for every agent.
Anything MCP-compatible connects to the whole hub at once.
INTEGRATION DIRECTORY
Connected once. Available by scope.
AGENTS
- research-agent
- workplace-agent
- coding-agent
One endpointMCP · OPENAPI · DATABASECONNECTED ONCE
- Google Drive
Gmail- Slack
GitHub- Notion
Salesforce
Illustrative selection — every connection is granted per agent and per user.
your-agentclaude-codesupport-bot
⬡ Agentic FabriqOutlookGmailSlackGitHubyour APIPostgres
CONNECT YOUR AGENT ONCE — EVERY INTEGRATION FABRIQ HAS IS AVAILABLE TO IT
THREE LAYERS. ONE DECISION.
The boundary travels with the work.
Explore the context attached to every request.
What is this agent allowed to do?
research-agent has a defined purpose and a bounded set of tools.
Whose authority travels with it?
The request acts for researcher, using that person’s permissions.
Where does the action land?
The policy applies to the requested action within Connected tools, before the tool executes.
QUESTIONS
A closer look.
Do we rebuild integrations we already have?+
No. Agents call tools through Fabriq’s endpoint and inherit the connections your organization has already made — connect once, centrally.
What about systems that aren’t on the shelf?+
Import an OpenAPI spec, register an MCP server, or connect Postgres behind guardrails — and internal systems stay behind your firewall.
Who manages all those tool credentials?+
Fabriq does — connections live in the vault, tokens refresh automatically before expiry, and disconnecting a tool cuts access everywhere at once.
Who decides which agents get which tools?+
You do — every grant is per agent and per user, and agents only ever see the tools they’re allowed to use.
How do agents connect to the hub?+
Anything MCP-compatible points at one governed endpoint and sees every tool it has been granted — no per-tool client code to write or maintain.
Can we manage integrations for our customers too?+
Yes — the same hub works across external users: your product’s agents act with each customer’s own connections and consent, managed per user. That mode ships on the Growth plan.