OSuite OSuite.ai
Sign in Request access
Runtime governance for AI agents

Approve high-risk
AI work
before it runs.

OSuite is one governed path for runtime-connected AI actions — review before launch, keep replay-ready evidence after execution, and produce buyer-ready proof across every tool and runtime.

Model-agnostic Azure, AWS or on-prem No agent code changes
{{ heroDot }} Governed action
{{ heroStatusLabel }}
{{ governViz }}
claude-code · lane: coding-hook HIGH RISK
$ rm -rf ./build && deploy --env=prod {{ heroCmdEdge }}
certificate ac_8f2a authority deployer externality external
{{ heroMark }}
{{ heroResultTitle }}
{{ heroResultSub }}
signing decision receipt…
receipt{{ heroReceiptId }}
decision byyou · operator
state{{ heroStateLabel }}
{{ heroPolicyNote }}
One layer across
Codex Claude Code MCP Framework SDKs API gateways Browser automation Managed agents
Built by Ond

At Ond, we build governance infrastructure so AI agents can serve organizations without escaping human authority.

OSuite is Ond's first flagship product: a runtime control plane for the actions agents take inside real businesses. The mission is not to slow AI down. It is to make authority, evidence, and blast radius visible before autonomous work touches production systems.

01 / Authority

Keep the deployer in control of actions inside its own environment.

02 / Evidence

Make every consequential action explainable, bounded, and replayable.

Read the Ond mission →
The thesis

Oversight is not
a control.

If a system can’t say whether an action was stopped, merely reviewed, or reconstructed after the fact, it isn’t governing that action — it’s narrating it. Governance becomes credible only when it is encoded as runtime structure, closure semantics, and portable proof.

— Oversight Is Not a Control, Ond Holdings, 2026
The method

Route. Review. Prove. On one governed object.

01

Route

Every action becomes one canonical, hashable object — the same whether it surfaces as a shell command, an SDK call or a hosted session.

02

Review

High-risk actions route to a human before they execute. Approval binds to the action itself — not raw text a rewrite could slip past.

03

Prove

Each decision closes into a replay-ready proof bundle: signed receipts that survive export, audit and runtime change.

04

Control

One authority layer across every runtime lane — coding hooks, gateways, managed agents and observer-only integrations.

The console

From intercepted to accountable, on one object.

One pane of glass for security, procurement, legal and platform teams.

studio.osuite.ai/reviews
Governance / {{ demoCrumb }}
{{ demoTitle }}
Search actions
{{ streamDot }} 3 awaiting
One action, three runtime forms
coding-hook
$ rm -rf ./build && deploy --env=prod
framework sdk
client.deploy(env="prod", purge=True)
api gateway
POST /v1/deploy  {"env":"prod"}
Canonical action
publish.external
fingerprintca_7b2c…0af
authoritydeployer
enforceableyes
APPROVAL BINDS HERE
AgentActionRiskDecision
claude-code deploy --env=prod HIGH
mcp:stripe refund charge ch_3Q… MED
codex read package.json LOW auto-approved
Routing signals
Blast radius9 / 10
Externality · destination8 / 10
Reversibility2 / 10
8.6
composite risk
REQUIRES APPROVAL
Proof bundle · replay-ready · buyer-exportable
12:04:24deniedac_8f2a · publish.externalj.okafor
12:01:50approved7b2c…0af · refund.issuepolicy:auto
11:58:02approvedc19d…44b · identity.changem.chen
11:52:31approvedf80a…1c2 · data.exportpolicy:auto
Get started · Connect

Run one command. First governed action in minutes.

You shouldn’t have to learn platform vocabulary before you can use it. Run the installer from the project root, finish the browser handoff, and send your first governed action — OSuite detects the runtime lane for you.

bash — ~/your-repo
curl -fsSL https://studio.osuite.ai/install.sh | bash
→  detecting agent runtime…
✓  detected: claude-code · coding-hook lane
→  opening browser handoff…
✓  workspace confirmed · terminal approved
✓  connected — send your first governed action 

Run it inside the repo root where the agent actually works. Force a runtime lane in advanced setup only if auto-detection is wrong.

1
One-command installer
Run it from the project root. The installer auto-detects Claude, Codex, SDK or MCP-style lanes.
2
Browser login handoff
The terminal prints a handoff URL. Confirm the workspace, then approve the terminal connection.
3
API key fallback
If handoff is unavailable, copy .env.example to .env, paste OSUITE_API_KEY — and don’t commit it.
4
Try one governed action
Ask the agent to run one low-risk action, then confirm OSuite saw it in Reviews or Decisions.
Golden path
Connect agent → Trigger risky action → Review in OSuite → Continue execution
Research at Ond

OSuite is a deployed implementation of published research.

External field signal
Security, platform, and governance reviewers keep asking for the same thing: turn agent oversight into a runtime control they can approve, explain, and audit.

Featured research
arXiv

PCAA: Proof-Carrying Agent Actions

A deployer-side governance primitive for binding approval to the action an agent will actually execute, rather than to a mutable prompt, wrapper call, or after-the-fact log.

Read the research →
Open PCAA article → View all research →
Built for the moments that need control

For the people accountable when an agent acts.

01 Launch approval Hold a deploy, an outbound publish or a destructive command for human review before it executes. → 02 Buyer-ready proof Export a signed trust packet that shows a customer or auditor exactly what your agents did — and who approved it. → 03 Runtime governance Apply one policy across coding hooks, MCP tools, SDKs, gateways and managed agents — without touching agent code. → 04 Post-launch monitoring Keep watching connected actions after rollout, with evidence that closes on a single governed object. →

Questions, answered.

{{ f.a }}