You own what your agents do.

Set boundaries on what AI agents can do. Enforce policy before execution and retain the evidence.

Partner program

See agent activity and policy outcomes

Loading product preview

labour value a year, modelled
$227K
compliance time returned a year, modelled
2,275 hrs
assumed automation of evidence work
35%
The problem

Autonomy needs more than oversight.

  1. No one can review every action.
  2. Rules live in documents, not in the path of the action.
  3. Evidence is rebuilt by hand, weeks later.
Who we are

The people building Gentrail.

We are a five-person team with experience across Amazon, defence systems, financial technology, and applied AI. We bring engineering, design, and business leadership together to make autonomous software accountable.

Meet the team
Fit with your environment

Runs in your AWS account.

Keep your deployment in your AWS account. Observe through existing traces; enforce controls with the Gentrail SDK or a tool hook.

Integration details Setup guide

Observe

  • OpenTelemetry
  • AWS Bedrock
  • OpenAI Agents SDK
  • LangChain
  • LangGraph
  • Vercel AI SDK
  • Strands

Control

  • Gentrail SDK
  • Claude Code
  • Codex

Clouds

  • AWS
  • Azure · planned
  • Google Cloud · planned
Join our partner program

Define what safe autonomy means for your business.

Bring a real agent workflow. Together, we will define the boundaries and the evidence you need to evaluate it.

Prove a real use case.

A focused partner program for security and AI platform leaders, with agreed scope and success criteria.

  • Runtime control, compliance, or both
  • Agreed success criteria
  • Hands-on support

Next: we reply within 2 - 3 business days.

Partner program

Start free

Monitoring, detection, and tracing in your AWS account.

  • 60 days, then read-only
  • Up to 10 agents
  • AWS bills cloud costs separately

Next: get a key, install, connect an agent.

Start free
News and updates

Latest from Gentrail

All updates
PRODUCT

Connect. Control. Certify.

From discovery to enforcement to audit-ready evidence. Manage your full compliance process from one platform.

app.gentrail.io
Gentrail trace explorer
DISCOVER
Full visibility at scale
Agent inventory · Shadow detection · Behaviour
CONTROL
Runtime enforcement
Block · Redact · Allow · Incident alerts
CERTIFY
Compliance evidence
Decision trace · Audit reports · Frameworks
SECURITY
Your data stays yours
Encrypted · Isolated · Flexible deployment
01
monitored shadow unowned
DISCOVER

Full visibility, no matter how many agents you run

Connect your AWS account and Gentrail maps every agent automatically. No manual inventory. Starting with AWS.

Built for scale and complexity

Whether you have fifty agents or fifty thousand, Gentrail discovers and catalogues them across environments without manual effort.

Shadow agents surfaced automatically

Agents deployed outside of governance, running in production, accessing data, making decisions with no oversight. Gentrail finds them.

Continuous behaviour tracking

What data each agent accesses, which APIs it calls, what actions it takes. A living behavioural profile that grows with every interaction.

02
AGENT action request POLICY ENGINE identity scope limit exceeded no approval pre-execution PASS BLOCK → Email · Dashboard EVIDENCE LOGGED per decision
CONTROL

Enforce the rules, before anything goes wrong

Every agent action is evaluated against your policies before it executes. Bring your own rules or start with built-in framework coverage.

Real-time enforcement, not retroactive alerts

Block, redact, or allow at the decision boundary, before the action executes. Violations are caught and stopped before they reach production.

Bring your own policies or start covered

Upload compliance documents in plain language. Or select from built-in frameworks and let Gentrail generate enforceable rules automatically.

Policy editor for custom rules

Write policies in the dashboard. Built-in framework packs come pre-loaded so you start with coverage. Add or override rules in the editor.

Incidents reach you instantly

Blocked actions, policy violations, and anomalies reported through email, the dashboard, or webhooks.

03
team A team B ... FINANCE NETWORK DATA ENG SOC 2 Type II 98% covered NIST AI RMF 1.0 94% covered ISO 42001:2023 91% covered EVIDENCE GENERATED signed · tamper-evident · audit-ready
CERTIFY

Every decision leaves proof.

Tamper-evident records generated automatically for every policy decision. Run compliance reports on demand. Hold evidence for as long as you need.

Full decision trace per agent

Who asked, what happened, which policy applied, what evidence was produced. Every decision is searchable, filterable, and exportable.

Compliance reports on demand

Run a report for any framework, any agent, any time window. Evidence is mapped to controls automatically. No manual assembly required.

Reports where your team works

Weekly posture reports, executive summaries, and incident digests delivered via email or the dashboard.

SECURITY & DEPLOYMENT

Your data stays yours. Always.

Gentrail runs in your AWS account. Your agent traces and audit evidence never leave your infrastructure. Encrypted at rest and in transit. Never used to train any model. No call-home - licensing is verified offline, inside your account.
Encrypted in transit and at rest
RBAC / Roles
Data residency controls
No customer data for training
INTEGRATES IN ABOUT 15 MINUTES

Add the Gentrail SDK to your code.

from aigentrail import GentrailGovernanceHook
from strands import Agent
agent = Agent(
hooks=[GentrailGovernanceHook()],
...
)
Python and Go SDKs available.
SUPPORTED FRAMEWORKS
SOC 2 Type II
NIST AI RMF 1.0
NIST AI RMF GenAI Extensions
ISO/IEC 42001:2023
ISO/IEC 27001:2022
HIPAA · coming soon
+ more coming
Upload compliance documents and Gentrail generates enforceable rules. Or select from built-in framework coverage and start governed immediately.

Ready to see it?

The live demo replays real agent incidents against the actual product. When you want it on your own agents, download the free tier or book a call.

See the live demo Download free tier Book a call
PRICING

Start in your cloud. Upgrade when you're ready.

Free for 60 days in your AWS account. Upgrade to Unlimited when you need compliance.

Free

The real product in your AWS account. Not a hosted sandbox.
$0 for 60 days
ScopeMonitor and detect
ScaleUp to 10 agents
Time60-day evaluation, then read-only. About 30 days of trace history.
Download
Available on GitHub. Deploy to your AWS in about 15 minutes.
INCLUDES
  • See every agent running in your AWS
  • Trace / invocation viewer
  • Alerts via email and dashboard
  • One-click AWS (Bedrock) connect
  • Python SDK
  • API keys
  • Create custom rules and apply compliance frameworks
  • Real-time violation detection
  • Violation triage
  • Tool catalog
  • Self-hosted BYOC (AWS) deployment
  • Login / onboarding

Unlimited

Govern, audit, and prove it.
Custom pricing
ScopeFull compliance and governance
ScaleUnlimited agents
TimeContinuous, full retention
Book a call
Or use the contact form below.
EVERYTHING IN FREE, PLUS
  • Compliance dashboard
  • Built-in frameworks (SOC 2 Type II, NIST AI RMF 1.0, NIST AI RMF GenAI Extensions, ISO/IEC 42001:2023, ISO/IEC 27001:2022)
  • AWS security scan (66 checks)
  • Audit periods
  • Auditor workpapers
  • Signed attestations
  • Sample reviews
  • Sign-off workflow
  • Drift alerts
  • Printable compliance report
  • Crosswalk views
  • Aux evidence records (CUECs, subprocessors, data-flow)
  • Agent classifications
  • Policy agent scoping
  • Inbox
  • Org chart
  • Roles
  • PII redactionComing soon
  • Decision ReplayComing soon
  • AI policy extractionComing soon
  • Sync enforcement APIComing soon
Discover

Automatic discovery of your Bedrock and SDK agents. Live inventory, full trace and tool-call visibility.

Trust

Your traces and evidence stay in your cloud - licensing is verified offline, with no call-home. Encrypted at rest and in transit.

Upgrade in place

When you need audit evidence, compliance frameworks, and unlimited agents, swap the license. Same install, same data.

FREQUENTLY ASKED
What is the 60-day evaluation?

Free runs the real product in your AWS account for 60 days. Full monitoring, detection, and tracing for up to 10 agents. After 60 days it goes read-only. Your data is preserved.

How do I upgrade?

License swap in place. Same install, same data, same AWS account. Compliance dashboards, audit periods, attestations, and unlimited agents unlock immediately.

What happens after 60 days?

The install stays. Read-only access continues so your team can browse the inventory and historical traces. Capture pauses until you upgrade.

Where does my data live?

In your AWS account. Gentrail is BYOC only. Encrypted at rest and in transit. Your agent traces and audit evidence never leave your infrastructure, and licensing is verified offline with no call-home. Optional AI-assisted features (policy rule extraction, tool classification) call third-party model APIs using your own API keys and provider accounts - Gentrail provides no model service, those calls never reach us, and the features stay off until you configure them. Never your traces, and never to train Gentrail models. If you opt into cross-account compliance scans, you grant a read-only, ExternalId-gated role that you can revoke at any time.

How long does setup take?

Cloud scanning starts within minutes of connecting your AWS account. Deploying Gentrail into your own AWS account (BYOC) takes about 15 minutes with our Helm and CloudFormation templates. Adding the SDK for runtime enforcement on an agent typically takes under an hour.

Do you use customer data for training?

No. Never. Your traces, policies, and evidence belong to you and are not used to train any model, ours or anyone else's.

What compliance frameworks are supported?

SOC 2 Type II, NIST AI RMF 1.0, NIST AI RMF GenAI Extensions, ISO/IEC 42001:2023, and ISO/IEC 27001:2022 are supported today, with more on the way.

How do I get pricing for Unlimited?

Talk to our sales team. Pricing depends on agent count and compliance scope. The contact form below gets you a same-week response.

Let's talk about your use case.

Tell us about your setup and we'll come back with a plan that fits.

Thank you.

We'll be in touch within one business day.

ABOUT GENTRAIL

Built by operators of the systems now being governed.

We've built production-grade AI at Amazon, designed enterprise policy and authorization engines, and run financial-grade infrastructure at the scale where a governance failure becomes a business risk.

THE MOMENT WE KNEW
WHAT WE BUILT
Agentic AI systems, policy engines, financial infrastructure, at enterprise scale.
WHAT WE SAW
Agents making decisions no one could trace. Compliance teams scrambling weeks after incidents.
WHAT WAS MISSING
A governance layer that works at runtime, not after the damage is done.
Dale Thoma
Founder · Product

Dale Thoma

Dale leads product at Gentrail, drawing on more than 20 years across Amazon, startups, and operational defence systems. That includes 12+ years at Amazon, with experience in large-scale financial systems, engineering leadership, and applied science, and 7+ years building defence systems.

His experience spans financial operations where correctness and compliance are essential, and defence systems where operational reliability, safety, and certification requirements shape how software is built and run. This work requires translating standards into controls, verifying system behaviour, and maintaining evidence that requirements are met.

That experience shapes Gentrail’s approach: define what an AI agent is permitted to do, enforce those boundaries, and preserve evidence of its actions.

View Dale’s LinkedIn profile
WHY WE BUILT GENTRAIL

The gap between agent deployment and agent governance is already costing companies millions.

Enterprises are deploying AI agents at unprecedented speed. But governance infrastructure hasn't kept up. The tools that exist were built for a world where humans make decisions.

Agents change that equation. They act autonomously, chain decisions, and access sensitive data without asking. Someone had to build the control layer. So we did.

WHAT WE BELIEVE

Three convictions that shape the product.

01

Governance at runtime, not after the fact

Retroactive audits don't prevent harm. Policy evaluation has to happen at the moment an agent acts, before the damage is done.

02

Evidence is a byproduct, not a project

Compliance evidence should be generated automatically from enforcement, not assembled manually by teams months later under audit pressure.

03

Speed and safety aren't tradeoffs

Policy evaluation happens in the execution path, at the moment of action, so agents stay fast while staying governed. You shouldn't have to choose between shipping and compliance.

THE TEAM

Built by operators, not observers.

Trust is becoming the premium in AI. As agents move from suggesting to acting, the hard part isn't the model. It's runtime control, policy that holds at the moment of action, and audit-ready evidence afterward. Trust, but verify. It's a runtime systems problem before it's a compliance one.

We've built exactly those systems: production-grade AI at Amazon, AI/ML research in models, enterprise policy and authorization engines, and financial-grade pipelines processing billions of transactions a day. High-scale, high-integrity work where traceability and correctness are non-negotiable. Across 30+ combined years at Amazon and decades more across financial services, cloud, and regulated industries, including $1.5B+ in P&L. We built Gentrail because we've watched autonomous software outrun the controls around it.

Start a conversation.

Whether you're exploring agent governance or ready to move, we'd like to hear what you're working on.

Thank you.

We'll be in touch within one business day.

FREE BYOC EVALUATION

Generate your free Gentrail license

60-day evaluation · up to 10 agents · runs in your own cloud. Your key appears here and we email a copy.