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

# AI Overview

> Connect AI agents and MCP clients to Bronto, run AI-driven investigations, and turn raw telemetry into structured, queryable observability data.

Bronto gives you multiple ways to bring AI into your observability workflows — and to observe the AI your own applications run.

Whether you want to connect an MCP client, run the server locally, investigate production issues with AI, instrument an LLM-powered application, or automatically structure raw telemetry, this section will help you get started.

<CardGroup cols={2}>
  <Card title="Hosted MCP" icon="cloud" href="/ai-features/hosted-mcp">
    Connect Claude or another MCP-compatible client directly to Bronto’s hosted endpoint. No local server required.
  </Card>

  <Card title="Bronto Agent Skills" icon="plug" href="/ai-features/bronto-agent-skills">
    Install Bronto's Cursor plugin or Claude Code plugin workflows so coding agents understand Bronto MCP and logging tasks.
  </Card>

  <Card title="Local MCP" icon="terminal" href="/ai-features/local-mcp">
    Run the Bronto MCP server on your own machine with Python and connect it to Claude Code or another compatible agent.
  </Card>

  <Card title="BrontoScope" icon="sparkles" href="/ai-features/brontoscope">
    Automatically investigate events and issues with AI and get scope, likely causes, next steps, and supporting evidence in seconds.
  </Card>

  <Card title="AI Investigation Reports" icon="file-lines" href="/ai-features/ai-investigation-reports">
    Attach AI-generated investigation reports to monitor alerts using a prompt that tells Bronto what to check.
  </Card>

  <Card title="LLM Observability" icon="brain" href="/ai-features/llm-observability">
    Capture prompts, responses, token usage, and model metadata from your own LLM-powered applications using the OpenTelemetry GenAI semantic conventions, and search them in Bronto.
  </Card>

  <Card title="Custom Parser" icon="wand-sparkles" href="/core-features/custom-parser">
    Use AI to turn unstructured telemetry into structured, queryable events with custom parsers tailored to your data.
  </Card>

  <Card title="Vibe Building" icon="bolt" href="/ai-features/vibe-building">
    Build custom observability interfaces with tools like Lovable and v0 on top of Bronto APIs.
  </Card>
</CardGroup>

## What you can do

<CardGroup cols={2}>
  <Card title="Connect AI to your data">
    Give agents access to datasets, keys, field values, search, and aggregated analysis workflows through MCP.
  </Card>

  <Card title="Investigate incidents faster">
    Use BrontoScope to automatically assess impact, identify likely causes, and surface recommended next steps.
  </Card>

  <Card title="Enrich monitor alerts">
    Add AI Investigation Reports to monitors so alerts arrive with automated analysis and recommended follow-up.
  </Card>

  <Card title="Reduce setup friction">
    Start quickly with Hosted MCP, or choose Local MCP when you need more control over how the server runs.
  </Card>

  <Card title="Observe your own LLM apps">
    Instrument applications that call OpenAI, Anthropic, Amazon Bedrock, or LangChain with the OpenTelemetry GenAI semantic conventions and send that telemetry to Bronto.
  </Card>

  <Card title="Make unstructured data usable">
    Create custom parsers that extract fields from application, system, and custom data formats so your data is easier to search and analyze.
  </Card>

  <Card title="Build focused custom UIs">
    Use Bronto APIs to create workflow-specific dashboards, service maps, and operational views.
  </Card>
</CardGroup>

## Recommended workflows

### 1. Connect an agent to Bronto

Start with one of the MCP options:

* Choose **Hosted MCP** for the simplest setup
* Choose **Local MCP** if you want to run the server yourself

<CardGroup cols={2}>
  <Card title="Hosted MCP" href="/ai-features/hosted-mcp" icon="cloud">
    Managed by Bronto. Best for the fastest path to an MCP connection.
  </Card>

  <Card title="Local MCP" href="/ai-features/local-mcp" icon="terminal">
    Runs locally with Python. Best when you need control or a self-managed setup.
  </Card>
</CardGroup>

### 2. Investigate a production issue

When you hit an unfamiliar issue, use BrontoScope to get an immediate starting point for your investigation.

<Card title="Investigate with BrontoScope" href="/ai-features/brontoscope" icon="sparkles">
  Ideal for quickly understanding impact, likely causes, and the next queries to run.
</Card>

### 3. Add AI analysis to monitor alerts

If you want investigations to run automatically when alerts fire, configure AI Investigation Reports on the monitor itself.

<Card title="AI Investigation Reports" href="/ai-features/ai-investigation-reports" icon="file-lines">
  Best for monitor-driven workflows where responders need a diagnosis attached to the alert.
</Card>

### 4. Improve the quality of your data

If your data is hard to search because it is unstructured, create a parser first.

<Card title="Structure data with Custom Parser" href="/core-features/custom-parser" icon="wand-sparkles">
  Transform raw telemetry into structured fields that are easier to search, group, and analyze.
</Card>

### 5. Instrument an LLM-powered application

If your own application calls an LLM — directly, via Amazon Bedrock/AgentCore, or through LangChain — capture that as GenAI telemetry rather than plain logs.

<Card title="LLM Observability" href="/ai-features/llm-observability" icon="brain">
  Best for teams shipping LLM or agent features who want prompts, responses, token usage, and model metadata searchable in Bronto.
</Card>