phronesis

AI-native advisory practice for the Broomva ecosystem. Encodes top-firm consulting methodology (Three Horizons, MIT CISR digital maturity, JTBD, Value Proposition Canvas, RICE/ICE/WSJF, NPV + Real Opt

SKILL.md

GitHub

phronesis — AI-native Advisory Practice

Status: Phase 1 complete (v0.1.2-pre) — substrate + 15 active frameworks + 5 stages + 7 deliverables + L1–L5 linter + M7 extraction pipeline. Phase 2 engagement-driven (M4 stage CLI + M5 autonomous-mode stub-to-real path). Track progress: docs/methodology.md (canonical reference), docs/adr/0001-0008, and the design spec at ~/broomva/docs/superpowers/specs/2026-05-06-phronesis-design.md.

What it does

Runs AI/digital consulting engagements through 5 stages:

  1. Intake — stakeholder map, interview guide, strategic thesis (kills "hagamos algo de IA" — Bision empirical 100% failure)
  2. Maturity Scan — MIT CISR + Gartner AI dimensions, gap analysis, capability heatmap
  3. Use-Case Ideation — JTBD + VPC, with IdeationSource diversity check
  4. Prioritization — RICE/ICE/WSJF, ROI Model with sensitivity, Impact-Effort Matrix
  5. Roadmap Synthesis — Three Horizons, QuantumBlack ML lifecycle pilots, baselines captured

Each stage has a non-negotiable human-review gate.

When to invoke this skill

  • User asks for AI/digital consulting work — maturity assessment, roadmap, use-case prioritization
  • User mentions Spanish-language equivalents: "consultoría", "evaluación de madurez", "hoja de ruta"
  • User describes a Fast Discovery, discovery sprint, or strategic-thesis-finding engagement
  • User wants to extract reusable IP from a completed engagement (cross-engagement learning)

When NOT to invoke this skill

  • User wants generic "do AI for me" without scope — phronesis explicitly refuses to start without a strategic thesis
  • User wants implementation work, not advisory — phronesis produces plans and recommendations, not running systems
  • User wants change-management or organizational design — that's D-scope (deferred to a later phase)
  • User wants only personal-decision support — use strategy-skills instead

Commands (CLI mode — most land in M4)

phronesis init <tenant> --industry <industry> --region <iso> --sponsor "Name (Role)" --scope "..."
phronesis intake --interview "..." --transcript path/to/file.md
phronesis intake --thesis "..."         # captures StrategicThesis (L1)
phronesis review intake --approve | --revise
phronesis scan --frameworks mit-cisr-digital,gartner-ai
phronesis ideate --frameworks jobs-to-be-done,value-prop-canvas
phronesis prioritize --frameworks rice --roi unit-economics,real-options
phronesis roadmap --frameworks three-horizons --pilot-design quantumblack-ml
phronesis render --all
phronesis lint                          # P3, P7, P8, L1-L5 enforcement (blocking)
phronesis status
phronesis bookkeep <slug>               # extract anonymized learnings to research/entities/

Five C-level failure modes structurally prevented

Bision-observed Frequency Linter rule Type primitive
Sin tesis estratégica 100% L1 STRATEGIC_THESIS_REQUIRED StrategicThesis
Casos mal priorizados 87% L2 DIVERSE_IDEATION_SOURCES IdeationSource
Datos no preparados 74% L3 DATA_READINESS_GATE DataReadinessAssessment
Desconexión negocio-tech 61% L4 ADOPTION_METRIC_REQUIRED AdoptionMetric
Sin medición de ROI 48% L5 BASELINE_REQUIRED BaselineSection

Documentation

  • Methodology: docs/methodology.md — canonical methodology reference (10 sections: stages, deliverables, frameworks, gates, linter, anonymization, modes, discovery loop, IP extraction)
  • ADRs: docs/adr/0001- through 0008-
  • Design spec: ~/broomva/docs/superpowers/specs/2026-05-06-phronesis-design.md
  • Linear: Phronesis project
  • Quickstart: (M4 — pending stage CLI commands) docs/quickstart.md