The Research Patterns
Swipe to show menu
Core Concept
Research at work is not academic research. It's the "I need to make a decision and I don't have time to read forty browser tabs" kind. This chapter gives you three recipes so that when you sit down to research anything, you know exactly which shape you're building and how to brief Cowork.
How It Works (Step-by-Step)
Step 1 - Recognize the three shapes
- Scan. Broad survey. Lay of the land.
- Compare. Options side by side on the dimensions that matter.
- Brief. Decision-ready recommendation.
Scan and compare aren't standalone deliverables - they're the inputs. The thing you hand to your team, or save to act on later, is always a brief.
Step 2 - Learn the SCAN recipe
Four steps.
- FRAME - Question, scope, and depth.
- SEARCH - Parallel searches across web and connectors.
- SYNTHESIZE - Cluster findings. Note where sources disagree.
- SURFACE - Overview, organized, with citations on every claim.
Depth in the brief equals depth out of the scan. Be specific about how deep to go.
Step 3 - Learn the COMPARE recipe
Four steps.
- SET CRITERIA - Name what matters before you start looking.
- GATHER - Evidence per option, per criterion.
- SCORE - Where each wins. Where the tradeoffs are.
- TABULATE - Clean table, citation in every cell.
You set the criteria. Cowork can't guess your weighting.
Step 4 - Learn the BRIEF recipe
Four steps.
- QUESTION - The decision in one sentence. Not the topic - the decision.
- EVIDENCE - Pull in the scan, the compare, anything relevant.
- RECOMMEND - Take a position. With reasoning.
- CAVEATS - What would change the call. What's still uncertain.
This is the only one of the three that ships.
Step 5 - Learn the three universal rules
- Cite as you go. Attach the source the moment you make a claim, not after.
- Verify what matters. Numbers, names, dates, claims with stakes - spot-check.
- Structure first. Section headers before any prose. Then fill the skeleton.
Why This Matters
Without a framework, research tasks bloat into meandering essays and never land as decisions. With these three recipes, every research request becomes a shaped, executable brief, and every deliverable ends with a real position taken.
What Should Be Possible After This
- Name the three research patterns and identify which one fits a given task.
- Recite the four steps of each recipe.
- State the three universal rules that hold across all three.
- Recognize that scan and compare feed the brief, and the brief is what ships.
Thanks for your feedback!
Ask AI
Ask AI
Ask anything or try one of the suggested questions to begin our chat