KLA Control Plane
Control what your agents can do in your systems.
KLA Control Plane connects your operating rules to the tools your agents use. It applies the policy, routes exceptions to a person and records what happened.
Your execution path
An agent requests an action.
Send a payment, update a customer record or call an external service.
KLA Runtime or a connected checkpoint in your existing stack
Evaluate identity, action and context.
- Allow / Warn
- Can proceed
- Require approval
- Hold for review
- Block
- Stop the call
Your connected systems
Execute only when permitted.
The tool receives the permitted request and returns its result or failure.
Payments, case management, data services and business applications
Correlate the captured request, policy decision, human review and tool result. Coverage depends on the paths and sources you connect.
Choose where to begin.
Govern. Operate. Assure. Prove. Open the product area that matches your team’s next job.
Govern
Agent Studio and Library
Build each agent in one place, with its instructions, tools, knowledge and rules. Changes ship as a reviewed agent release. Rules that several agents share live in Library.
Policy decisionsOperate
Agents and Decision Desk
Follow governed execution and resolve the actions that need an authorized human decision.
Human oversightAssure
Lineage Explorer and Assurance Center
Investigate captured execution, review Findings and track Remediation Plans as operating behavior changes.
Execution LineageProve
Audit Trail and Control Mapping
Review the recorded history of governed actions, map it to your controls and export scoped evidence.
Control mappingFit the control to your architecture.
Run the agent through KLA, or keep your stack and add KLA checkpoints around its tools. Both paths produce the same decisions, approvals and evidence.
Run through KLA
Use KLA Runtime for governed execution, with policy decisions and approval holds connected to the run.
Explore the execution modelGovern in Place
Add KLA checkpoints around the tools in your existing agent stack with the Node.js or Python SDK.
Review integration optionsReview an example evidence package before defining the records your evaluation needs.
Inspect an Evidence Room sampleStart with the workflow you want to put into production.
Bring the agent, its tools and the rules it must follow. We map the integration and agree what the first workflow has to show.
A 30 minute working session to scope one workflow.