discover — what to build next

← Book index · grounded feature proposals & tool evaluation

What it is: the pre-DEFINE research skill. It answers "what should we build?" and "should we adopt X?" with cited evidence and measured usage — never invented competitors, prices, or numbers.

How it works

flowchart LR
  A[read the product<br/>as it is] --> B[market scan<br/>web, cited, today's date]
  B --> C[measure usage<br/>writes read-only queries itself]
  C --> D[gap + architecture<br/>feasibility check]
  D --> E[ranked proposals<br/>anchored effort] --> F[you pick → define]

Best cases

• Roadmap planning — quarterly "what's next", table-stakes vs differentiators.
• Tool / platform evaluation — pros/cons, current pricing, license limits, integration fit, lock-in (e.g. OpenStreetMap → Google Maps).
• "Are our features enough?" — though whole-app health judging belongs to assess; discover leads on net-new.

Examples

itqan:discover "our invoicing SaaS has 200 users — what should we build next?"
itqan:discover "we use OpenStreetMap — evaluate switching to Google Maps: pros/cons, pricing, fit"

What you get

discovery.md: current state → cited market scan (checked at today's date) → usage/goal context (measured where possible, asked where not) → ranked proposals, each with evidence, target first-users, effort anchored to the modules it touches, and a validation path (MVP/prototype).

Hand-offs

Chosen proposals flow to define. Adopted tools get an ADR and construct's dependency-adoption checks (license, maintenance health). assess's "next moves" feed in as input.

Pro tip: have rough usage numbers handy (users, feature usage) — it will measure what it can from your own schema and ask only for the rest; ranking quality tracks data quality.