Insights

Pulling signal out of the data you already have. Agents read across your CRM, product analytics, finance system, support queue, and internal docs, then synthesize the parts nobody has time to stitch together by hand.

The point of an insights agent is narrative, not dashboards. If you already have a dashboard, you don't need another one. You need someone to read the dashboard and tell you what it means.

What people build here

• Weekly business review. Every Monday morning, the agent pulls the metrics that matter (revenue, pipeline, product engagement, support volume, top accounts) and writes a short narrative for the leadership thread. Numbers with context, not a number dump.
• Customer health scoring. Combines product usage, support tickets, NPS, and CSM notes into a single per-account health signal. Flags accounts trending the wrong way before renewal time, with the reason and a suggested next step.
• Customer feedback synthesis. Reads every support ticket, product review, sales call, and NPS response from the week. Groups by theme. Produces the weekly digest your product team will actually read instead of skim.
• Competitive monitoring. Watches competitors' release notes, pricing pages, job postings, and social presence. Summarizes what changed and what it means for your roadmap or positioning.
• Metric drilldown. When a key metric moves (up or down), the agent slices it (geography, plan, cohort, channel, feature flag) and reports back with the likely cause and the evidence.
• Board prep. The week before each board meeting, the agent drafts the metrics section and the qualitative narrative, pulling from your CRM, accounting system, and product analytics. You edit and ship.
• Pre-meeting briefings. Before a meeting with a specific customer, you get a one-pager: account history, recent usage, open tickets, last few touches, anything worth knowing.

How to ask for one

Every Monday at 6 AM Pacific: pull last week's revenue and active-customer count from Stripe, weekly active users and top-3 feature usage from PostHog, and open ticket count from Linear. Compare to the prior week and the four-week trailing average. Write a 250-word summary covering what's up, what's down, and what's worth poking at this week. Post to #leadership-weekly with the numbers in a small table below the narrative.

Connections you'll usually want

Your CRM, your product analytics, your finance or billing system, your support tool, your docs and meeting notes, and whatever specific data sources the people reading your reports actually care about.