Building Witness, layer 1 of 5

Every agent,
governed.

One neutral layer that audits and controls your AI agents — any model, any framework, any cloud.

LangChain · CrewAI · AutoGen · Raw APIs — registered · logged · governed

Governing agents built onLangChainCrewAIAutoGenRaw API

Built an agent in a week. Security said no. It's still in pilot.

Run agentic AI pilots
79% of enterprises run agentic AI pilots1
Run agents in production
11% run them in production2

The blocker isn't the model. It's governance.

Sources: 1 PwC AI Agent Survey, 2025 · 2 Deloitte Tech Trends 2026

not yet adoptedstuck in pilotin production

§ 01

First cohort · Witness

See everything

Connect any agent in one line. Instence registers it and starts a tamper-evident audit trail.

agent.py
refund-agentCrewAIOpenAIhigh risk
Records stream into the trail events logged

§ 02

Next · Gatekeeper

Control what matters

One policy for every provider. Each action is allowed, alerted, escalated, or blocked — in real time. Escalations land in Slack for one-click human approval.

refund-agent CrewAIOpenAIhigh risk refunds $1,240
Instence escalates.

escalateRefunds over $1,000 need a person to approve.Approved via Slack · 38 ms · logged

§ 03

Later · Liftoff & Scale

Prove it to auditors

The trail becomes evidence. EU AI Act, NIST AI RMF, ISO 42001 — reports assemble themselves. Ship through staging gates with one-click rollback.

PSrefund-agentto productionRejectApproved

Evidence assembled

EU AI ActNIST AI RMFISO 42001

No model to sell you.

OpenAI, Anthropic, Mistral, local — all at the same distance. Switch models, keep your controls.

Roadmap

We're building layer one.

Every layer after it runs on the record it keeps. The first cohort gets it first.

  1. First cohort

    Layer 1

    Witness

    Register every agent and keep a tamper-evident trail of what it does.

  2. Next

    Layer 2

    Gatekeeper

    Policies on every action, and human approval where it counts.

  3. Later

    Layer 3

    Liftoff

    Staging gates, rollback and cost control on the way to production.

  4. Later

    Layer 4

    Scale

    Teams, enterprise governance and audit-ready reports.

  5. Later

    Layer 5

    Conductor

    Agents working together, still governed.

Join the first cohort, and shape Gatekeeper with us. Request early access

Get your agents to production.

Agent count
Framework (optional)

  1. 01

    Tell us about your agents

    Your stack, and what's blocking production.

  2. 02

    We onboard in small cohorts

    Hands-on setup.

  3. 03

    Watch your inbox

    We reach out with your slot.