AI Business Designer — Skills

🌐 Browse the pack online

A skills pack for Claude Code and Cowork that helps business designers, consultants, and entrepreneurs do higher-quality business planning, analysis, and problem definition — spotting and sizing opportunities, building a decision-ready business case, prioritizing AI initiatives, and framing a demo without overpromising — instead of ad hoc prompting. Taken as a whole, the pack also works as a map of the competence field around AI-assisted business design: what the discipline actually consists of, broken into named, independently usable pieces.

9 core packs · 4 populated specialisation packs · 111 skills · 4 audit agents · self-contained (no required external services) · CI-validated · MIT license

New here? QUICKSTART.md gets you from zero to your first skill run in about five minutes.

Why this exists

Most "business strategy" prompting is either too generic to be useful or too confident about numbers nobody actually verified. This pack tries to fix both: each skill is a narrow, named technique anchored to a public framework or method (cited in the skill itself), and every output is explicitly framed as a decision-support draft, not a finished decision — assumptions are marked, not hidden.

On depth — read this before you assume more than is here. The pack includes a number of heuristics and decision rules, drawn from general practice and from the owner's own experience. It does not contain deep proprietary playbooks, hard-won pricing/negotiation heuristics, or strong, opinionated value judgments about how business should be done — the kind of thing a consultancy keeps for paying clients. In that sense this pack is intentionally open and general-purpose rather than a proprietary methodology in disguise.

Nothing here is speculative. Every technique in this pack — at every maturity level — is grounded in either the owner's own real consulting engagements or an established, professionally used framework (cited in the skill itself). There is nothing invented or experimental: applied with sound judgment, in the right context and to a fitting client situation, it works. The maturity marking below tells you something narrower than "does this work" — it tells you whether the owner personally has run that specific skill through a real engagement yet, or whether it's still a well-grounded framework awaiting that. Both are real and usable; the marking exists so you always know which one you're looking at. See "How maturity is tracked" below.

When to use this — and when not to

Use it when... Don't reach for it when...
You're in Claude Code / Cowork doing business opportunity, business case, AI-strategy, or change-communication work and want a structured technique instead of a blank page You need legal, tax, regulatory, or financial advice — that requires a licensed professional, not a skills pack
You want to know how confident to be in a given technique before you rely on it You want a single number or decision handed to you with no visible reasoning — that's not what this produces, by design
You want something that works offline, with your own numbers, no external accounts required Your task is software engineering — this pack is about business design, not code

Quick start

/plugin marketplace add Pilot2Service/AI-Business-Designer
/plugin

The first command registers this repository's .claude-plugin/marketplace.json catalog (marketplace name: ai-business-designer-skills); the second opens the interactive menu — choose Browse and install plugins, pick the packs you need (e.g. opportunity-recognition, business-case-and-analysis), and install. Full walkthrough, including what a good first run looks like: QUICKSTART.md.

What's inside

Nine core packs cover the parts of AI-assisted business design that come up across most engagements; four specialisation packs go deeper into a specific situation (research commercialisation, AI-native startup design, Business Model Canvas facilitation, public-sector AI service design). Read the two tables below as much as a map of the discipline as a list of what to install — click a pack name to see its actual skill-by-skill list (every skill in this repo, named and described in one line, lives in that pack's own README.md; nothing here is just a number).

Core packs

Pack Helps you... Skills
strategic-thinking break down a fuzzy problem into a testable hypothesis (MECE / issue trees) 6
opportunity-recognition scan, evaluate, size, and write up a business opportunity 8
business-case-and-analysis build an ROI/NPV business case with risks and assumptions made explicit 6
ai-strategy-and-governance prioritize AI use cases, scope a PoC, and check responsible-AI readiness — including a premise check for whether an idea is real reshuffle or just automation 15
change-and-communication plan change management and executive communication — including closing the information gap with a decision-maker before pitching, and stripping AI jargon before it ships 6
business-design-frameworks apply classic business/value-modeling frameworks (value chain, strategy canvas, category design) — an intentionally growing collection 6
prototyping-and-demonstration frame and deliver a credible demo or PoC without creating false expectations of production-readiness 5
data-strategy-and-literacy diagnose data's role in a business and read data critically before trusting it 6
human-ai-collaboration-design design and audit intentional human-AI oversight — HITL maturity and confidence routing, AI behavioral/accuracy specification, override-rate auditing, protecting expert agency, and the individual practitioner's own judgment discipline 6

Specialisation packs

Pack What it gives you Skills
research-commercialisation A structured path from a research result to a commercialisation decision: IP disclosure and ownership checks, route selection (licensing vs. spin-out), funding-option mapping, and team/equity questions, plus a working self-assessment tool 12
ai-native-startup-design A founder-facing path from a raw AI opportunity to a buildable spec: customer understanding (ICP/JTBD), RICE-scored MVP selection, tiny-core feature-freeze discipline, a PRD an AI coding agent can act on, human-oversight design for the resulting product, and a tool-stack decision 9
business-model-canvas Facilitation-grade Business Model Canvas work: running the session, diagnosing a finished canvas for weak spots, correcting common misunderstandings about what a canvas is, matching a model against a library of known innovation patterns, a set of resilience/risk heuristics (economics prototyping, organizational-resistance mapping, channel economics, operational risk scanning), and two AI-native facilitation techniques (AI-drafted starting canvas, AI-scaled customer interviewing) 19
public-sector-ai-service-design A public-sector lens on top of the core AI-strategy, business-case, and stakeholder skills: screening AI ideas for public value and mandate fit, mapping the distinct stakeholder types and veto points in public organizations, navigating procurement and public funding, regulatory and equity guardrails for citizen-facing AI, and a six-element decision-readiness model for public decision bodies 7

How maturity is tracked

Every skill's frontmatter only ever contains name and description — no confidence claims live there. Internally, skills_index.json tracks two things for every skill as the owner's own refinement backlog: maturity (scaffold → draft → validated → canonical) and source_layer (research = built from a public framework, owner = converted from the owner's own field-tested experience). This isn't a completeness rating surfaced in the pack's own documentation — every skill, at every level, is grounded in either a named professional framework or the owner's real practice, never invented. Details: meta/maturity_levels.md.

Delegatable agents

Four packs include a read-only subagent (agents/*.md) that you can invoke separately (via the Task tool) to stress-test a skill's output before it goes to a decision-maker — a second, independent pass, not a rubber stamp.

Agent Pack What it checks
assumption-stress-tester business-case-and-analysis Adversarially challenges a business case's assumptions before the number goes to leadership
market-sizing-cross-validator opportunity-recognition Cross-checks a TAM/SAM/SOM calculation with an independent top-down/bottom-up method
competitive-blind-spot-scanner business-design-frameworks Looks for un-scanned competitors or angles in a competitive/positioning analysis
ai-initiative-readiness-auditor ai-strategy-and-governance Audits an AI initiative's scoring and governance checklist for gaps before approval

None of these agents edit anything — each returns a findings table for a human to act on.

Optional external data (not a dependency)

This pack never requires an external service to function — every skill works with numbers you provide, and marks assumptions explicitly when you don't have one. If your environment happens to have a relevant data MCP connected (e.g. a market-sizing data source), the market-sizing-tam-sam-som skill and its cross-validator agent can use it instead of, or to cross-check, an assumption. See meta/external-data-mcp.md for the (unaudited, third-party) candidates considered.

Quality assurance

python3 scripts/generate_index.py   # rebuilds skills_index.json from disk
python3 scripts/validate.py         # checks structure + frontmatter

.github/workflows/validate.yml runs both automatically on every push and pull request, so a broken frontmatter or an out-of-sync index blocks the merge without anyone having to remember to check by hand.

A note on language

The skill instructions themselves (SKILL.md, CLAUDE.md, reference files) are written in English, so the pack is usable and reviewable without a Finnish-language dependency. Regardless of the instruction language, Claude will respond to you in whichever language you use.

Contributing

See CONTRIBUTING.md for how to add a skill, fill in a Refinement notes section, or raise a skill's maturity level once it's actually been used.

ai-business-designer-skills/
├── README.md
├── QUICKSTART.md                      start here
├── AGENT_GUIDE.md                     how an agent should use this pack
├── CONTRIBUTING.md
├── CHANGELOG.md
├── LICENSE
├── skills_index.json                  machine-readable index (generated — don't hand-edit)
├── .claude-plugin/
│   └── marketplace.json               lists every installable pack
├── .github/workflows/
│   └── validate.yml                   CI: generate_index.py + validate.py on every push
├── scripts/
│   ├── generate_index.py
│   └── validate.py
├── meta/
│   ├── repo_purpose.md
│   ├── skill_design_principles.md
│   ├── frontmatter_schema.md          name + description only — nothing else allowed
│   ├── maturity_levels.md
│   ├── competency_map.md
│   ├── shared-guardrails.md           single source for the disclaimers every pack shares
│   └── external-data-mcp.md
├── strategic-thinking/                [plugin] 6 skills
│   ├── .claude-plugin/plugin.json
│   ├── CLAUDE.md                      pack-wide guardrails (a safety net, not the main mechanism)
│   ├── README.md
│   ├── skills/<skill-id>/SKILL.md
│   └── references/
├── opportunity-recognition/           [plugin] 8 skills + agents/market-sizing-cross-validator.md
├── business-case-and-analysis/        [plugin] 6 skills + agents/assumption-stress-tester.md
├── ai-strategy-and-governance/        [plugin] 15 skills + agents/ai-initiative-readiness-auditor.md
├── change-and-communication/          [plugin] 6 skills
├── business-design-frameworks/        [plugin] 6 skills + agents/competitive-blind-spot-scanner.md
├── prototyping-and-demonstration/     [plugin] 5 skills
├── data-strategy-and-literacy/        [plugin] 6 skills
├── human-ai-collaboration-design/     [plugin] 6 skills
├── specialisation-packs/
│   ├── ai-native-startup-design/       9 skills
│   ├── business-model-canvas/          19 skills
│   ├── public-sector-ai-service-design/ 7 skills
│   └── research-commercialisation/     12 skills
├── templates/
│   ├── skill-template/SKILL.md
│   └── specialisation-pack-template/README.md
└── playbooks/                         pre-built skill chains for common tasks
    ├── idea-to-decision.md
    └── ai-initiative-scoping.md

About this project

This repository serves two purposes at once. First, and primarily, it's a working tool — skills meant to actually be installed and used. Second, it's a deliberate demonstration piece: of what a modern AI Business Designer's skill set looks like broken down into its component parts, and of the ability to design, structure, and maintain a production-quality, skills-based AI toolkit — the context-engineering decisions (maturity tracking instead of unverifiable confidence claims, a single source for shared guardrails instead of drift across a dozen copies, delegatable audit agents instead of a single pass of self-review, CI validation instead of an honor system) that make a toolkit like this trustworthy rather than just extensive. If you're evaluating this repository rather than using it, that structure is the part worth looking at closely.

License & author

MIT — see LICENSE.

Author: Tommi Järvinen · Adductor Magnus Oy.