Give your teams the freedom to use AI agents, with policies that protect business data and controls that keep risky actions in check.
Built on UC San Diego
Read why Two Small FishSet policies, see agent activity, and investigate risks across employee devices, business applications, and cloud workspaces.
An agent may read a document, write code, and update an application to finish one request. Keep the work connected to the same approved task and company policies.
Use approved information
Draft, calculate, and write code
Stay within the task’s limits
Check consequential changes
An agent’s change of approach should not become a change of permission.
Keep the agents your teams rely on. Bring their work into one security program, with policies and visibility across supported integrations.
Available protections vary by integration and deployment. We’ll confirm the right coverage for your workflow. Product names and logos identify their respective owners.
Give people room to get useful work done with the agents they already use. Keep approved tasks moving, protect sensitive information, and explain when an action needs review.
See the task behind an alert and the activity that led to it. Review related evidence together, set consistent policies, and focus attention on the decisions that matter.
Built for operations-heavy businesses, from business services and logistics to internal manufacturing teams. Operations, IT, AI platform, and security share the outcome.
Resolve order and service exceptions without widening access or making an unrelated change.
Explore the scenario ↗02Turn approved data into useful analysis while governing code, exports, and proposed updates.
Explore the scenario ↗03Reconcile records and prepare corrections while keeping payment and recipient changes bounded.
Explore the scenario ↗04Delegate document and application work with clear file, account, and publishing boundaries.
Explore the scenario ↗05Explore fixes, inspect dependencies, and validate the exact artifacts promoted from an isolated workspace.
Explore the scenario ↗06Keep retries and delegated work inside the original authority, even after hours of execution.
Explore the scenario ↗See how a security team reviews an unexpected finance-file change, checks the evidence, and brings the employee into the conversation.
Scroll to follow the investigation
1
2
3
01 / Activity
See what the agent did and the outcome it reported.
A finance-file change appears alongside the team’s ordinary project work.
See which details came from the agent and which were independently observed.
1
2
3
02 / Policy & alert
See which company policy applies and why the action needs attention.
Inspect the reported write to forecast.csv and its source context.
Carry the finding and its supporting evidence into an investigation.
1
2
3
03 / Investigation
Keep the original alert and related activity together.
Review the resource, reported outcome, and matched condition together.
Preserve a record of changes and assessments as the investigation develops.
1
2
3
04 / Employee context
Ask the employee about the task and the change under review.
Retain the response alongside the investigation evidence.
Document the decision. An explanation does not automatically change policy.
Actual product · synthetic demonstration. The console and evidence workflow are real; source reports and the employee explanation are simulated. These screens demonstrate investigation, not preventive blocking.
Silent walkthrough with on-screen captions. The annotated screens above provide a text alternative for the workflow.
Published test frameworks help us evaluate protection and detection. In our internal ADR-Bench evaluation, Gensee improved recall and accuracy while reducing false positives.
92.9%RecallBaseline 73.8%
11.5%False-positive rateBaseline 13.8%
89.1%AccuracyBaseline 84.5%
Gensee’s internal evaluation on Uber ADR-Bench. Results are specific to the evaluated configuration.
Baseline and Gensee results across three evaluated threat categories. Higher is better.
+18.8 percentage points
+34.6 percentage points
+15.7 percentage points
Gensee’s internal evaluation using an adapted AgentCanary harness; these are not official leaderboard results.
We achieve higher recall and accuracy with a lower false-positive rate on the Uber ADR-Bench dataset.
| Metric | Baseline | With Gensee | Change |
|---|---|---|---|
| Recall Higher is better | 73.8% | 92.9% | +19.1 pp |
| False-positive rate Lower is better | 13.8% | 11.5% | −2.3 pp |
| Accuracy Higher is better | 84.5% | 89.1% | +4.6 pp |
Recall measures how many positive cases are detected. False-positive rate measures how often negative cases are incorrectly flagged. “pp” means percentage points. Baseline refers to the configuration recorded in Gensee’s internal evaluation, not Uber’s production system.
0.6%–1.2%runtime overhead
10–400 msper request
Know what agents are doing, protect sensitive work, and give your security team the context to act.
See it in your workflowReview agent activity and the files, tools, and applications involved in a task.
Set policies for sensitive data and business actions. Apply safeguards before risky work goes further.
Surface suspicious behavior with the context of the task, so teams can assess the activity behind an alert.
Bring related activity, policy decisions, and employee context together in one investigation.
Founded by UC San Diego professor Yiying Zhang and former Google engineering leader Shengqi Zhu, Gensee brings systems research to agent safety. Meet the team
How live workspace branches return useful results—and where their boundaries end.
Read the research ↗ LONG-HORIZON AGENT SAFETYWhat real incidents teach us about shared services, persistent state, and control over time.
Read the analysis ↗Answers for the people doing the work—and the people protecting it.
Gensee Crate Enterprise helps companies use AI agents safely. It brings company policies, agent activity, safeguards, and investigations together across employee devices, business applications, and cloud workspaces.
Gensee connects suspicious agent activity with the task, relevant policies, and supporting evidence. Security teams can review related findings together, understand what happened, and decide how to respond.
Teams with agents that already read sensitive information, run code, or change business systems. Operations and support are a natural starting point, with analytics, finance, and engineering workflows also in scope. The operational owner, platform or IT team, and security team define the boundaries together.
Gensee works with supported coding agents, automation agents, and cloud platforms. These include Claude Code, Codex, Antigravity, Cursor, GitHub Copilot, Claude Cowork, ChatGPT Work, n8n, OpenClaw, Hermes Agent, Omnigent, Databricks Agent Bricks, Glean, Workday, Atlassian Rovo, AWS Bedrock, and Azure AI Foundry. We confirm available protections for your integrations and deployment.
Gensee supports employee devices running macOS, Windows, or Linux, agents inside business applications, and managed cloud workspaces. Policies, activity review, and investigations come together in a central dashboard.
Approved tasks can proceed under company policies. People are brought in when an action needs attention, so teams can keep routine work moving and review consequential exceptions.
Gensee can inspect code, packages, and tool activity, flag risks, and help teams choose a safer path. Available safeguards depend on the connected agent and deployment.
No. Recovery applies to supported local state. It cannot unsend an email, reverse an arbitrary remote transaction, or recall disclosed data. Those effects need controls at the application or resource boundary before they occur.
Start with one workflow. We’ll confirm the supported agents and applications, agree on policies, and demonstrate the controls and evidence available in that setup.
Bring the task your team delegates.
We’ll help map the boundaries that matter.
Start with the work you already delegate.
See what it takes to protect the path from request to result.
