Trust
Ship work, not surprises
Every external send, write, or purchase gets canonicalized, hashed, and queued for human review. Your team sees exactly what an agent will do — before it does it.
AgentHub MVP — public beta
AI agents your compliance team will sign off on.
A marketplace for human-approved, evidence-cited, audit-ready agents that plug into the tools your team already uses. No black-box autonomy. Every external action is reviewable before it runs.
- Hash-pinned approvals on every external write
- Span-level citations on every answer
- Tenant-isolated retrieval out of the box
- Eval-gated releases — no agent ships unverified
Built for trust before logos · controls included from day one
Hash-pinned approvals
Span-level citations
Tenant-isolated retrieval
Tamper-evident audit chain
Eval-gated releases
What you get
Outcomes, not infrastructure.
AgentHub is opinionated where it matters: trust, evidence, and the boring guarantees that turn a demo into something a real team can run.
Evidence
Answers you can defend in a review
Replies and reports come back with span-level citations: source, page, and the exact passage. No more chasing down where the model got that number.
Use cases
Built for teams who own outcomes.
Every pack ships with approval policies, fallback modes, and an eval suite — so the on-call shift never has to wonder what the agent did last night.
Document triage
Turn contracts, policies, and meeting transcripts into queryable knowledge with citations on every answer.
Support escalations
Draft, classify, and route tickets without ever sending a message your team didn't approve.
Marketplace
Featured agent packs.
Each pack ships with eval scores, approval policies, and a sandbox you can try before you wire it into a real workspace.
How it works
Four steps. No surprises.
From browsing to first run, every step is reviewable, reversible, and auditable.
- Step 01
Pick a pack
Browse the marketplace, read the eval report, scan the approval policy and connector scopes.
- Step 02
Try the sandbox
Drive the agent against synthetic data. See its tool calls, citations, and approval prompts before installing.
- Step 03
Install with policy
Pick the workspace, set the approval policy, and authorize the connectors your team already uses.
Pricing
Pay only for what your agents ship.
Predictable seat pricing, transparent run metering, and no hidden infra markup.
Starter
$0
Free during beta
For teams kicking the tires.
- 1 workspace, up to 3 seats
- Document triage pack included
- 100 runs / month
- Approval + audit baseline
Most popular
Team
Ship agents your compliance team can sign off on.
Start with a single pack, install in under five minutes, and turn on more capabilities as your trust grows.
No credit card required during beta. Cancel anytime.