> ## Documentation Index > Fetch the complete documentation index at: https://docs.resolve.ai/llms.txt > Use this file to discover all available pages before exploring further. # Tuning Accuracy **Guided Tour · Part 5 of 5** [Resolve Overview](/resolve-overview) → [Connect an Integration](/connect-an-integration) → [Team Knowledge](/D5xH-team-knowledge) → [Auto Investigations](/auto-investigations) → Tuning Accuracy Improving investigation accuracy is an **iterative loop**: give feedback, find the gap, update knowledge or integrations, then re-run and compare. This final part of the Guided Tour walks through that loop. ## Start with an investigation whose root cause you know Pick an investigation where you're **certain of the real root cause** — ideally an older [auto-investigated](/auto-investigations) alert you've already diagnosed. If you don't know the true answer, you can't tell whether tuning moved Resolve closer to it. For example, a *CPU Utilization High* alert you've seen before. ## Step 1: Give feedback Open the investigation report and use the **thumbs up / thumbs down** control: * If the [root cause](/resolve-overview) matches what you expected, give it a **thumbs up** and note why. * If it's wrong, give it a **thumbs down** and describe the problem — for example, *"the root cause looks incorrect because…"*. Feedback does two things: it records the outcome so Resolve can learn from it, and it lets you **track this investigation** as you iterate. All feedback is visible later under **Analytics → Feedback**, where you can see every thumbs-up and thumbs-down across your investigations. ## Step 2: Find the gap When an investigation misses, there are usually only a few causes: * **A missing integration** — Resolve didn't have access to a data source it needed. Review what's connected in [Integrations](/integrations) and add what's missing via [Connect an Integration](/connect-an-integration). * **A bad or mis-scoped query** — for example, Resolve queried the wrong log index. * **Missing guidance** — Resolve needs more direction on how to investigate this alert or how to use an integration. ## Step 3: Update knowledge or integrations Go to the relevant team's [Team Knowledge](/D5xH-team-knowledge) and fill the gap: * Add a **sample log query** or note the correct index so Resolve queries the right place. See [Alert Runbooks](/knowledge-setup-guide). * Add or refine **dashboard guidance**, **Docs**, or a **[Skill](/skills)** that encodes the right procedure. * If a data source was missing, connect it — see [Setup & Integrations](/setup-and-integrations). The full menu of what you can adjust is in the [Knowledge Setup Guide](/knowledge-setup-guide). ## Step 4: Re-investigate and compare Back on the investigation report, open the **⋯** menu and choose **Reinvestigate alert** (use the default option). Resolve re-runs the **same alert** with your updated knowledge and integrations, so you can compare the new result against the original. * Got a better outcome? Give it a **thumbs up** to lock in the win. * Still off? Thumbs down, refine again, and re-run. Repeat this loop — feedback → find the gap → update → re-investigate — and accuracy climbs steadily over time. This is the same loop that keeps Resolve improving long after onboarding. *** ## You've completed the Guided Tour 🎉 You now know how to interact with Resolve, teach it your systems, automate investigations, and tune them for accuracy. Where to go next: * Wire up the rest of your stack in [Setup & Integrations](/setup-and-integrations). * Explore the full [Features](/features) set, including [Mitigation Actions](/mitigation-actions) and [Tasks](/tasks). * Build on Resolve with the [Developers](/developers) tools — [plugins](/ai-assistant-plugins), the [MCP server](/resolve-mcp-server), and the [REST API](/resolve-rest-api). **Previous:** [Part 4 — Auto Investigations](/auto-investigations)