Let Agents take action
Extend Agents with Tool Apps, MCP servers, Skills, and A2A Remote Agents. Call limits, timeouts, and human confirmation are controlled at runtime.
Workrun
A local-first workspace for AI automation
- Local-firstYour workspace stays on your machine
- Checkpoint resumePause for human review, then continue
- Evaluation-readyValidate changes before publishing
REAL PRODUCT SURFACEBring judgment and delivery onto one canvas.
Order cancellation: risk routing and regression evaluationCANVAS · RUN HISTORY · EVALUATION
Workflow editor in action: agents, conditional routing, and human review collaborate on one canvas.
WORKFLOW-NATIVE · LOCAL-FIRST · READY TO SHIP
More than orchestration. Automation that can deliver.
Section titled “More than orchestration. Automation that can deliver.”Put Agents, Python, MCP tools, and human judgment into a readable workflow. Every run leaves state, tool traces, and evaluation evidence behind. Workrun turns one useful AI collaboration into a capability your team can run again and keep improving.
Product media placeholder · video
homepage/workflow-studio.mp4(45–60 seconds)
Show a new workflow, Agent configuration, a run, and a Human Review pause and resume. Use a real window recording and hide keys, customer data, and internal URLs.
Configuration is not a black box
Section titled “Configuration is not a black box”The workflow structure below shows what should remain visible and inspectable: where a node reads data, who makes a decision, when a human takes over, and which model and tools a run used.
● Saved
Customer feedback triageDraft
STARTFeedback inputfeedback
AGENTClassify and recommendStructured output
IF / ELSEHigh risk?priority == high
HUMAN REVIEWConfirm replyWaiting for approval
ENDCompletefinal_state
Keep code deterministic
Every Python App is an independent local uv project. Keep data processing, integrations, and business rules in the codebase you know.
Bring quality into publishing
Use run records, OpenTelemetry traces, evaluation suites, version comparisons, and quality gates to know whether a change is actually better.
From idea to team delivery, in one workspace
Section titled “From idea to team delivery, in one workspace”| What you need to do | How Workrun helps |
|---|---|
| Compose capabilities | Connect Agents, Processes, branches, subworkflows, and human nodes on a canvas. |
| Run reliably | Isolated node state plus checkpoint recovery for pauses, failures, and background runs. |
| See what happened | Inspect model messages, tool calls, script logs, and traces node by node. |
| Iterate safely | Validate workflow changes with evaluation suites and quality gates before publishing. |