The fourth mode brings research material to life in a format that matches the purpose: a navigable brief, a short script for a stakeholder who will not read a deck, or rules for a queryable research agent after the project ends.
Usable handoffs, then the four-mode loop.
Speed still matters — but speed in service of understanding, at the right time and in the right format to act.
The four modes are a navigation system. Real projects loop. If a readout goes badly, you usually skipped Challenge or Immerse. If fieldwork feels random, you usually skipped the narrow after Widen.
Core principles
- Immerse when findings will die in an appendix, or another team must query the learning later.
- Pretty packaging of weak evidence is failure — flag fragile claims from the challenge pass.
- Agent rules: answer only from sources; cite; separate evidence from inference; never invent quotes.
- Loop exit checks: Widen → Keep/Park/Drop; Notice → themes+tensions; Challenge → one change; Immerse → audience can decide.
- Capstone: same live task through all four modes, then verify the three riskiest claims against primary sources.
Next skill: Building agents in Copilot
Check yourself
When has Immerse in context mode succeeded?
Immersion is not decoration. It is speed in service of understanding — the right format so someone can act, query, or challenge the work.
Do this in Copilot
Finish the capstone checklist on one real piece of work. Share the one-pager (or agent rules) with a colleague and ask what they would decide differently.
Paste this into Copilot Chat and work through it before moving on.
Actionable research brief
Turn these research notes into a one-page brief for [AUDIENCE] who must [DECISION / ACTION]: [PASTE CHALLENGED INSIGHTS + KEY EVIDENCE] Structure: - Situation (5 lines max) - What we heard (bullets with source tags) - Tensions we are not resolving yet - Implications for [DECISION] - What we still need to learn - Recommended next conversation (who / about what) Tone: plain, specific, no hype. Flag any claim that is still fragile from our challenge pass.
- Optional agent spec: system rules that refuse answers not in sources, cite every claim, and label recommendations as suggestions with assumptions.
Did you run this in Copilot? Mark complete when you have tried it.
RecordedNavigate: press j for next lesson,k for previous.