Deployment Insights

Deployment Insights is Devpilot's AI-generated overview of what is going well, what is slowing you down, and what you should do next. Instead of digging through deployment logs one app at a time, you get a prioritized list of observations for the whole workspace — along with a health score that summarizes deployment reliability at a glance.

What insights surface

Insights are generated by analyzing the last 30 days of deployment activity across your workspace. Each insight is a small, focused card with everything you need to act on it:

• Title and description — A plain-language summary of the observation.
• Category — The kind of insight, such as a recommendation or a warning about a pattern.
• Severity — info, warning, error, or critical, so the most urgent items rise to the top.
• Priority — A numeric rank used alongside severity to order the list.
• Action text and action type — The suggested next step, for example "Review failing hooks" or "Open deployment history".
• Health score delta — How the insight affects the workspace or app health score.
• Metadata snapshot — A snapshot of the aggregated numbers (total deployments, success rate, and so on) that the AI used when producing the insight, so you can see exactly what it was looking at.

Health scores

Alongside the insight list, Devpilot surfaces two health scores:

• Workspace health score — A rolling score that reflects deployment success rate, failure patterns, and recurrence across every app in the workspace.
• Per-app health scores — The same idea, broken down for each app so you can see which apps are stable and which are struggling. The app list is sortable by last deployment time or score, and paginated when you have many apps.

Health scores are designed to be a quick glance indicator. Use them to pick the app that needs attention first, then drill into the insights and deployment history for the details.

Where insights appear

• Workspace dashboard — A summary card shows the workspace health score and the top few insights.
• Insights page — The full list, filterable by app and by category. This is where you triage and dismiss items.
• App view — When you open an app, insights scoped to that app are shown inline alongside its deployment history.

Acting on an insight

Refreshing insights

Devpilot regenerates insights automatically in the background. If you have just made a big change — for example, fixed a recurring failure or reconfigured a pipeline — you can trigger a manual refresh from the Insights page. Manual refreshes are rate-limited to two per minute per workspace to avoid overloading the AI service.

How Devpilot avoids duplicate insights

Each insight is fingerprinted from its workspace, category, and title. If the AI produces the same observation again while an existing, active insight already exists, Devpilot silently skips the duplicate. This means:

• You will not see the same recommendation listed twice.
• Dismissed insights stay dismissed — the AI will not regenerate an identical one while the active copy exists.
• New, genuinely different observations still appear as new insights.

Some insights also have an expiration time. If an observation is only relevant for a short window (for example, a spike in failures in the last 24 hours), it will automatically drop off the active list once it expires.

Tips for getting the most out of insights

• Start with the top of the list. The list is already sorted by priority and severity — working top-down focuses your attention on the highest-impact items.
• Pair insights with deployment diagnosis. An insight tells you "this app fails often"; the deployment diagnosis on a specific failed deployment tells you exactly why it failed. Use them together.
• Dismiss what does not apply. Dismissing noise keeps the list focused on actionable items.
• Watch the health score trend. A single number does not tell the full story, but a declining score across weeks is usually a signal to invest in deployment reliability.

Next steps