Skip to content

Insight extraction: connected tools

This routine reads the tools you have connected and saves what bears on your missions as insights. Your team's own conversations, tickets and pipeline are where most useful signal lives, and none of it reaches the agent unless something reads it.

Why it exists

Most of what matters to a mission happens somewhere you were not looking: a prospect objection in a sales call, a complaint in a support ticket, a deal that stalled in the CRM, a decision made in a Slack thread you were not in. The daily plan is only as good as what it knows, and nobody has time to copy that signal into Questo by hand.

The extractor is the sensor. It does the reading so the plan starts from what actually happened rather than from your memory of it.

What it does

You get one extractor per connected tool. It is created when you connect the tool, is on from the start, and goes away when you disconnect. Connecting a tool is asking Questo to work with it, so the scan starts straight away rather than being a second switch to find. A workspace with Slack and Gmail connected has two.

Each run reads only its own tool, over a window that starts where the last one ended (the first run looks back seven days). It uses your active missions as a lens:

  • the places named in a mission's strategy under insight sources (channels, labels, subreddits, pipelines) are read first. They decide where to look, never what to capture;
  • the latest mission update says what the plan is stuck on, so it looks for evidence that moves it, for or against;
  • what the mission shipped in the window is what it looks for reactions to.

It captures anything in the window that is:

  • progress, an obstacle or an opening against a mission's objective or a milestone;
  • customer or prospect signal: a need, an objection, an ask, praise, a reason someone churned;
  • pipeline movement: a meeting booked, a deal advancing or stalling;
  • a competitor or market move;
  • a decision, bet or learning the team commits to;
  • an internal change that shapes go-to-market, such as a product change or a blocker.

Pure noise (scheduling, greetings, reactions) and anything already captured is skipped.

What an insight looks like

Each one is a single specific claim of at most 140 characters, with a verbatim quote or link from the source as evidence. No evidence, no insight. It carries the day the thing happened, a link back to it when the source gives one, and the missions it bears on. Missions are tagged on evidence about the objective or a milestone, not because a signal shares a word with a mission's name.

When a signal is evidence about a competitor, one of its gaps or a persona, the insight is linked to that record and says whether it confirms or contradicts what the record claims. A prospect saying a competitor now has what a gap says they lack contradicts that gap.

What you see

Nothing arrives in your inbox. The extractor writes insights silently. You meet them in two places:

  • on the Insights page, where you can read, edit or delete them;
  • in the next morning's mission planning, where tagged insights are the primary new signal.

It reads and writes only insights. It never creates or edits missions, tasks or tenets, and it never asks you anything. An unattended run that stopped to ask a question would capture nothing.

When it runs

Weekdays, in the late afternoon, ahead of insight request, which waits for it.

Tuning it

  • Connect the right tools. An extractor only sees what its integration can read. See Slack and integrations for what each connection does and does not do.
  • Name your sources in the mission's strategy. Putting the channels, labels or pipelines that matter under insight sources is the single biggest improvement to what it finds.
  • Turn off one tool's extractor if its signal is mostly noise, without disconnecting the integration.