# Investigate and Fix Issues with Seer AI Debugging Agent > Use Seer Agent to ask what broke in plain language and get answers from your errors, traces, logs, and code in Sentry or Slack, then kick off an Autofix. **URL:** https://sentry.io/cookbook/seer-agent-use-cases/ --- Ask Seer Agent what broke and why in plain language, in Sentry or Slack, and it reasons over your errors, traces, logs, and code to find the root cause, then kicks off a fix. **Time:** 10–15 minutes | **Difficulty:** Beginner ## What You'll Learn - Enabled Seer agent for your organization - Use the agent to find the right telemetry data for specific areas of your application - Investigated errors and metrics to find the root cause of the issue - Identified the best person to help investigate an issue or action on a PR - Triggered an autofix using the Seer Agent **Topics:** Seer, Error Monitoring, Tracing, Logs, Issues ## Before you start **SDKs & packages** - [Sentry SDK](https://docs.sentry.io/platforms/) installed and reporting events (errors at minimum; traces and logs make answers sharper) - [Seer](https://docs.sentry.io/product/ai-in-sentry/seer/) enabled for your organization **Accounts & access** - [Sentry account](https://sentry.io/signup/) with at least one project - Open Team Membership enabled for your org (required for Seer Agent today) - Optional: [Slack integration](https://docs.sentry.io/integrations/notification-incidents/slack/) for Seer Agent in Slack **Knowledge** - Basic familiarity with Sentry issues and projects - A production question you actually care about answering ## Steps ### 1. Get started with Seer Agent Seer Agent is currently in beta and available to all users at no charge. To get started, just click the 'Ask Seer' button in your navigation bar. The fastest way to your team access to the Sentry Agent though is to use the [Slack integration](https://docs.sentry.io/integrations/notification-incidents/slack/#seer-agent). Using this, you can communicate with Seer Agent through direct messages, easily share issue and application diagnostics, and integrate the Seer agent into automated debugging workflows. To enable Seer agent in Slack, go to [Settings → Integrations → Slack](https://sentry.io/orgredirect/organizations/:orgslug/settings/integrations/slack) and install/update the app. [Seer Agent overview](https://docs.sentry.io/product/ai-in-sentry/seer/#seer-agent) ### 2. Learn all about your application Click **Ask Seer** on any page in Sentry to start a conversation. You don't need to pre-build a Discover query or pick the right explorer—describe what you're seeing in plain language. Good first prompts look like: - **Error spike** — "Why did checkout errors jump in the last hour?" - **Latency** — "What's causing p95 on `/api/orders` to climb since the last release?" - **User impact** — "Which issues are affecting enterprise customers today?" - **Correlation** — "Are these frontend timeouts related to a backend span failure?" [How Seer Agent uses telemetry](https://docs.sentry.io/product/ai-in-sentry/seer/#seer-agent) ### 3. Dig in on issues and share what's wrong Seer Agent connects the dots across everything you've sent to Sentry: [issue](https://docs.sentry.io/product/issues/) context, [traces](https://docs.sentry.io/concepts/key-terms/tracing/#whats-a-trace) and spans, [logs](https://docs.sentry.io/product/logs/), profiles, and linked code when your SCM integration is set up. Use it for investigations you'd otherwise bounce between explorers for: - **Root-cause hunting** — ask it to walk from a failing transaction to the span, log line, or stack frame that explains it - **Release diffs** — ask what changed after a deploy when error rates or latency moved - **Cross-service debugging** — ask whether a frontend symptom has a backend or dependency cause - **Follow-ups** — keep the same thread going as Seer reasons through evidence in real time [AI in Sentry capabilities](https://docs.sentry.io/product/ai-in-sentry/) ### 4. Find the right person for the fix Sometimes know who can help with an issue is just as important as knowing what exactly went wrong. You can use the Seer Agent to identify which of your teammates is best equipped to handle the fix based on previous commits. Seer agent will inspect the code history to find out when an issue started to emerge and then can assign the Sentry issue to that specific user. Looking to keep track of who's handling what? Here's an example flow you could create: - **Create an Issue** — Use the [Linear integration](https://docs.sentry.io/integrations/issue-tracking/linear/) to automatically open Linear issues - **Identify an issue owner** - Have the Seer Agent in Slack look at commit histories and identify the right user - **Assign them to the issue** — Use the Linear agent or use the new [Sentry Agent for Linear](https://docs.sentry.io/integrations/issue-tracking/sentry-linear-agent/) to have Seer generate a fix for you [Seer Agent sharing and history](https://docs.sentry.io/product/ai-in-sentry/seer/#seer-agent) ### 5. Hand the job off to Seer or your agents Once you know what the issue is and who the right person to assign the issue to is, it's time to fix it. Fortunately, it's never been easier to automate code fixes by using agents and the Seer agent works with a bunch of different agents. - **Seer Autofix** - You can trigger the [Seer autofix](https://docs.sentry.io/product/ai-in-sentry/seer/autofix/) workflow directly from the Seer agent in Slack with a single click. - **Agent Automations** - If you use tools like Claude routines or Cursor automations, you can [instruct agents to interact with the Seer agent in Slack](https://blog.sentry.io/automated-debugging-workflow-sentry/) to gather information about an issues and your application. [Autofix workflow](https://docs.sentry.io/product/ai-in-sentry/seer/autofix/) ## Summary **You ask. Seer digs.** Skip the explorer gymnastics—ask Seer Agent what broke, why it's slow, or what changed, and it reasons over your production data in real time. ## Pro Tips - Send more than bare errors—traces and logs are what let Seer Agent connect frontend symptoms to backend causes. - Start prompts with the symptom and a time window ("last hour", "since release X") so Seer scopes the search. - In Slack, drop the Sentry issue or trace link in the thread before asking; Seer Agent uses that context automatically. - Copy a strong Seer Agent conversation into your coding agent when the fix needs local repo context Seer can't see. ## Common Pitfalls - Seer Agent won't appear if Open Team Membership is disabled—it needs org-wide query access. - In Slack channels you must `@sentry` on every reply; forgetting the mention looks like the agent went silent. - Expect weaker answers when the project only reports exceptions with no tracing, logs, or code mappings. - Seer Agent is in open beta—treat outputs as an investigation aid and verify before you ship a fix. ## FAQ **How is Seer Agent different from Autofix?** Seer Agent is a conversational investigator: you ask questions and it searches across your telemetry. Autofix is a structured workflow on a specific issue—root cause, solution plan, and optionally a PR or coding-agent handoff. Use Agent to explore; use Autofix to remediate. **Does Seer Agent cost extra?** Seer is an add-on billed via active contributor pricing when Seer features are enabled on connected repos/projects. See Sentry's Seer pricing for current details. Seer Agent itself is in open beta. **Why don't I see Ask Seer?** Confirm generative AI / Seer is enabled in Settings → Seer, and that Open Team Membership is on. Seer Agent isn't available when Open Team Membership is disabled because it runs org-wide queries without finer-grained ACLs yet. **Can I use Seer Agent only in Slack?** Yes for day-to-day prompting, as long as Seer Agent is available to your org and Slack is connected. The Slack agent has the same capabilities as the web version and also reads thread context. You can still open Ask Seer in the Sentry UI anytime. **What data does Seer Agent look at?** Issue context, traces and spans, logs, profiles/performance metrics, linked GitHub or GitLab code, your guidance in the chat, and Sentry docs—whatever you've connected and sent to Sentry. ## Next Steps - [product/seer](https://sentry.io/product/seer/) - [Build a self-healing workflow with Seer](/cookbook/self-healing-workflow-seer/): Turn diagnosis into automation: Autofix root causes and open PRs or hand off to a coding agent. - [Catch bugs in PRs before they merge](/cookbook/ai-code-review-seer/): Have Seer review pull requests using real production error context. --- *Source: [sentry.io/cookbook/seer-agent-use-cases/](https://sentry.io/cookbook/seer-agent-use-cases/)*