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

# Honeycomb

> Connect Rootly AI to Honeycomb — traces, metrics, logs, BubbleUp investigations, Triggers, and SLOs — with a region selection and one OAuth authorization.

## Overview

The **Honeycomb** connector lets Rootly AI query your Honeycomb environments during an investigation. Honeycomb is where high-cardinality trace data lives, so this is the connector that answers *"which specific requests are failing, and what do they have in common?"* — the question raw dashboards usually can't.

Setup is a region selection followed by an OAuth authorization. There is no API key to create or paste.

<Info>
  Rootly AI's access is **read-only**. It authorizes with a read scope and can query your Honeycomb data but never create, modify, or delete Triggers, SLOs, or environments.
</Info>

***

## Before You Start

<ParamField path="Honeycomb Account" required>
  An account with access to the environments you want Rootly AI to query. Rootly AI inherits the read access of whoever authorizes the connection.
</ParamField>

<ParamField path="Your Honeycomb Region" required>
  Honeycomb runs separate US and EU instances and your data lives in exactly one. You'll pick this during setup, and picking the wrong one means Rootly AI authorizes against an instance that doesn't hold your environments.
</ParamField>

<ParamField path="Rootly Permission To Manage Connectors" required>
  Connecting a data source requires the permission to manage AI connectors on your Rootly account. If the **Connect** button is unavailable, ask a Rootly admin.
</ParamField>

<Tip>
  If you're not sure which region you're on, check the URL you use to sign in to Honeycomb. EU accounts sign in on a `eu1` host; US accounts don't.
</Tip>

***

## Connecting

<Steps>
  <Step title="Open the Honeycomb Card">
    Go to **Configuration → AI → Connectors** and click **Connect** on the **Honeycomb** card. The **Connect Honeycomb** dialog opens.
  </Step>

  <Step title="Choose Your Region">
    <ParamField path="Honeycomb region" type="US or EU" required>
      Pick **US** for environments hosted in the US region, or **EU** for environments hosted in the EU region. This is required — the dialog won't submit without a selection.
    </ParamField>
  </Step>

  <Step title="Authorize Rootly">
    Rootly redirects you to Honeycomb's authorization screen. Sign in if prompted and grant access. Rootly requests a read-only scope.
  </Step>

  <Step title="Confirm the Card Shows Connected">
    Back on the Connectors page, the Honeycomb card shows **Connected** with **Configure** and **Disconnect** actions. Rootly AI can now query Honeycomb during investigations.
  </Step>
</Steps>

***

## What Rootly AI Can Read

| Signal       | What it answers during an investigation                                                                                                      |
| ------------ | -------------------------------------------------------------------------------------------------------------------------------------------- |
| **Traces**   | Which requests are slow or failing, and where in the call path the time or error originates                                                  |
| **Metrics**  | Rate, error, and duration trends across the window around the incident                                                                       |
| **Logs**     | Individual events with their full attribute set, not just aggregate counts                                                                   |
| **BubbleUp** | What the failing requests have in common that the healthy ones don't — the specific customer, region, build, or endpoint driving the anomaly |
| **Triggers** | Which Honeycomb Triggers fired, and when, relative to the incident timeline                                                                  |
| **SLOs**     | Current SLO compliance and remaining error budget for the affected service                                                                   |

<Note>
  Rootly AI can read across the full range of Honeycomb queries rather than a fixed subset, and reaches for whichever read the current question needs.
</Note>

***

## During an Incident

*"Checkout latency is concentrated in one trace pattern. BubbleUp shows 94% of the slow requests share build `a3f21c` and the `eu-west-1` region — the other regions on the previous build look normal."*

Honeycomb earns its place in an investigation when the question is *which* requests are affected rather than *how many*. Error-rate dashboards tell you something is wrong; trace data plus BubbleUp tells you the failures all share one deploy, one region, or one customer tier — which is usually the fastest path to a cause.

***

## Best Practices

* **Connect Honeycomb alongside a deploy source.** Honeycomb tells you the failing requests share a build; [GitHub](/ai/connectors/github) tells you what was in that build. The pair answers "what changed" far faster than either alone.
* **Authorize with an account that can see every relevant environment.** Rootly AI inherits the authorizing user's read access. Authorize from an account scoped to one team's environment and investigations will silently miss the rest.
* **Keep your Triggers meaningful.** Rootly AI reads which Triggers fired and treats that as signal. Noisy Triggers that fire constantly reduce how much that signal is worth.
* **Reconnect after an account change.** The authorization is tied to the user who granted it. When that person leaves or their Honeycomb access is reduced, reconnect from another account rather than leaving a connection that quietly returns less data.

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="'Choose a Honeycomb region to connect'" icon="globe">
    The region field is required and the dialog rejects a submission without it. Select **US** or **EU** and submit again.
  </Accordion>

  <Accordion title="Connected successfully, but Rootly AI finds no data" icon="magnifying-glass">
    Almost always the wrong region. A US-region authorization against an EU account connects cleanly — the OAuth flow succeeds — but Rootly AI is then querying an instance that holds none of your environments. Disconnect, reconnect, and pick the other region. Check the host you use to sign in to Honeycomb if you're unsure.
  </Accordion>

  <Accordion title="Rootly AI sees some environments but not others" icon="key">
    Rootly AI inherits the read access of the account that authorized the connection. If that account can only see a subset of your Honeycomb environments, so can Rootly AI. Reconnect from an account with broader read access.
  </Accordion>

  <Accordion title="The authorization screen never opens" icon="triangle-exclamation">
    Usually a popup blocker or an existing Honeycomb session in a different account. Allow popups for Rootly, or sign out of Honeycomb in that browser and retry so you can authorize with the intended account.
  </Accordion>

  <Accordion title="Queries worked, then stopped returning results" icon="plug-circle-xmark">
    If someone revoked Rootly's authorization on the Honeycomb side, the card can still read **Connected** while every query fails. Disconnect and reconnect to establish a fresh authorization.
  </Accordion>
</AccordionGroup>

***

## Frequently Asked Questions

<AccordionGroup>
  <Accordion title="Do I need a Honeycomb API key?" icon="key">
    No. Setup is a region selection plus an OAuth authorization — there's no key to create, paste, or rotate.
  </Accordion>

  <Accordion title="Can Rootly AI change anything in Honeycomb?" icon="lock">
    No. The connector authorizes with a read-only scope. Rootly AI can query traces, metrics, logs, Triggers, and SLOs, but cannot create or modify Triggers, SLOs, or any other Honeycomb object.
  </Accordion>

  <Accordion title="Can I connect both a US and an EU Honeycomb instance?" icon="globe">
    One Honeycomb connection is active per team. If your data spans both regions, connect the one your incident-relevant environments live in and contact support about the other.
  </Accordion>

  <Accordion title="How is this different from connecting Datadog or Grafana?" icon="chart-line">
    They answer different questions. Datadog and Grafana are strongest on aggregate metrics and dashboards — *how bad is it, and since when*. Honeycomb is strongest on high-cardinality trace data — *which requests specifically, and what do they share*. Teams that run both connect both.
  </Accordion>

  <Accordion title="Does Rootly store our Honeycomb data?" icon="floppy-disk">
    Rootly doesn't keep a persistent copy of your Honeycomb data — queries run at investigation time and the response is used in reasoning only. Query results *do* appear in the LLM traces Rootly logs for quality monitoring, as with every tool call Rootly AI makes. See [Data Privacy for Rootly AI](/ai/data-privacy-for-rootly-ai) for the retention boundary.
  </Accordion>
</AccordionGroup>

***

## Related Pages

<CardGroup cols={3}>
  <Card title="Connectors Overview" icon="sparkles" href="/ai/connectors/overview">
    All connectors and how they fit together.
  </Card>

  <Card title="GitHub" icon="github" href="/ai/connectors/github">
    Pair with Honeycomb to tie a failing trace pattern back to the deploy that caused it.
  </Card>

  <Card title="Data Privacy for Rootly AI" icon="shield" href="/ai/data-privacy-for-rootly-ai">
    What Rootly AI sees, retention, and model training controls.
  </Card>
</CardGroup>