VibesFactory
Open console
AI agent infrastructure

The AI agent platform for production.

Build agents with saved versions, connect tools, and coordinate multi-agent workflows. Inspect every run in detailed traces. VibesFactory provides the infrastructure your agents need to run.

Open console Explore the platform
  • Host it yourself
  • Choose the version
  • See every run
WorkflowSteps run in order
Saved versionsRuns use the version you choose
Playground preview · illustrative conversation
Works with the tools you already use
AI model providersMCP toolsPostgreSQLOpenTelemetry
How it works

Build an agent, connect what it needs, and follow every run.

Each step builds on the last, so you can see how an agent is set up and what happened when it ran.

01

Create an agent

Write instructions and choose the model it should use.

Set up an agent
02

Add tools and data

Connect documents, memory, and tools the agent can use.

Connect resources
03

Build a workflow

Link agent and tool steps in the order they should run.

See workflows
04

Inspect a run

See the input, each step, the result, and any errors.

Explore run traces
Inside the platform

One place to set up agents, connect tools, and understand results.

Use a clear screen for each job: configure an agent, connect a workflow, or find out what happened in a run.

01 / BUILD

Set up an agent and publish a version.

Write instructions, choose a model, and add tools or knowledge. Publishing saves a fixed version for runs to use.

  • Edit instructions and model settings
  • Add tools, documents, and memory
  • Test the agent in the playground
02 / WORKFLOWS

Connect the steps in a visual workflow.

Start with an event, add agent or tool steps, and see how the pieces connect before you run them.

  • Arrange steps on a canvas
  • Send work between agents and tools
  • Pause for human review when needed
03 / INSPECT

See what happened in each run.

Follow the request through the agent, model, and tools. Open a step to see its input, output, status, or error.

  • Follow every step in one timeline
  • See results from tools and knowledge
  • Sensitive values are hidden in traces

Answer from your documents

Upload documents and let agents find useful passages with source citations.

KNOWLEDGE

Connect external tools

Add MCP servers and choose which tools your agents can use.

TOOLS · MCP

Set safety checks

Use guardrails to allow, block, or redact content, or request human review.

GUARDRAILS · APPROVALS
Coming next

More ways to test, launch, and track your agents.

These features are planned. They are not available in the console yet.

Planned

Evaluation

Test agents with example questions and compare versions to see which gives better answers.

COMPARE ANSWERS
Planned

Deployment

Let other apps call a chosen agent version through an API.

CONNECT OTHER APPS
Planned

Monitoring

Track successful runs, errors, speed, and cost after agents are live.

FOLLOW LIVE PERFORMANCE
Built to be clear and flexible

Know where your data goes and which agent version runs.

Host VibesFactory yourself. Keep workspaces separate. Run a specific published version, and inspect its results.

View the project on GitHub
Separate workspaces

Each workspace keeps its own agents, tools, and data.

Fixed agent versions

Each run uses the published version you selected.

Host it yourself

Run the platform on infrastructure you control.

Inspect each run

See how the agent reached its final response.

Ready to build?

Start with one agent.

Set it up, connect the tools it needs, and inspect its first run.

Open console