> ## Documentation Index
> Fetch the complete documentation index at: https://arize-ax.mintlify.site/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# AX Agent Improvement Loop

> Specialized workers that can investigate, run evals, triage security concerns, and even act as your SRE—in sandboxes with access to repos, skills, and full trace context from Arize.

<Info>
  AX Agent Improvement Loop is available in closed Enterprise beta. Contact your Arize account team for access.
</Info>

<Frame caption="The improvement loop: observe, surface, fix, and ship — with Signal, Agent Studio, Agent Swarms, and Agent Presets as entry points">
  <img src="https://storage.googleapis.com/arize-phoenix-assets/assets/images/arize-docs-images/agents/agent-improvement-loop.png" alt="Cycle diagram: traces land in Arize, Signal ranks issues, a sandbox worker drafts a fix, and you review and merge the PR so improvements flow back into the observed app" />
</Frame>

## Where to start

Use any combination—these are separate entry points, not steps in a pipeline.

* **[Signal](/ax/agents/get-started-with-signal)** — Turn on per LLM project. A scheduled worker scans traces and surfaces ranked issues with investigations.
* **[Agent Swarms](/ax/agents/manage-agents)** — Monitor active workers, browse sessions and automations, and open run detail. Start new work with **+ New Agent** → [Agent Studio](/ax/agents/agent-studio).
* **[Agent Presets](/ax/agents/agent-presets)** — Save reusable harness, sandbox, repo, and skill settings under **More → Agent Presets**.
* **How it works** — [Sandboxes, security, and data access](/ax/agents/how-agents-work); [skills and RBAC](/ax/agents/skills-and-permissions).

## Terms

* **Agent preset** — Saved harness, sandbox, optional repo, and skills. Managed under **More → Agent Presets**.
* **Session** — One interactive run from Agent Studio; follow the transcript and send follow-ups.
* **Automation** — Same worker configuration on a cron or metric trigger. Requires a preset.

## Related docs

* [Agent trajectory / tracing](/ax/instrument/agent-trajectory) — Instrument the app that produces traces
* [Agent Insights](/ax/observe/agent-insights) — Manual or Alyx-assisted analysis across traces
* [Alyx](/ax/alyx) — In-product copilot
* [Arize Skills](/ax/set-up-with-ai-assistants) — Run Arize workflows in your local coding agent

AX Agent Improvement Loop does not host your production-facing agent runtime. Your team runs workers against apps you already observe in Arize.