ai-opportunity-portfolio

Identifies, scores (5-dimensional model: Business Impact, Technical Feasibility, Data Readiness, Strategic Alignment, Speed to Value/Risk), and prioritizes AI use cases with a 2x2 matrix (Quick Wins /

SKILL.md

GitHub

AI Opportunity Portfolio

Purpose

Turns a raw list of AI use-case candidates (typically 20–100 items) into an objectively scored, prioritized portfolio, from which the 3–5 highest- value, lowest-risk items are selected to move forward. The skill deliberately separates two different opportunity types — incremental efficiency gains (making an existing process faster/ cheaper) and transformative innovation (new business that wasn't possible before current AI capabilities) — because they're assessed partly against different criteria.

Anchored in research

  • LinkedIn Skills on the Rise 2026 — AI Business Strategy
  • Market research: open "Senior AI Business Designer"-type job postings
  • A research report supplied by the user, "AI Business Designer in the Age of AI" (2026) — identifying AI opportunities at the strategic level (the original problem-type/data/flywheel/agentic-ness triage, now folded into point 4 below)
  • Research digest "Methods, Frameworks, and Competencies for Identifying AI Opportunities and Capacity in Business" (2026) — the 5-dimensional scoring model (a synthesis of several industry AI capability reports), the 2x2 prioritization matrix, the Value Play taxonomy for transformative opportunities, the Deploy-Reshape-Invent taxonomy

Method

  1. Assemble the raw list of candidates. Start from existing friction points and value-chain bottlenecks — not from technology. Two complementary ways to assemble the raw list:
    • Bottom-up (if the process is already precisely described): use ../task-level-decomposition-and-automation-fit/SKILL.md — its Automate/Augment-classified tasks are grouped here into larger opportunities.
    • Top-down (a fast first pass before a detailed process description): use ../ai-capability-pattern-matching/SKILL.md — it poses the client the diagnostic questions of a ready-made capability pattern library and produces a validated raw list. If neither has been used, collect the list directly from stakeholders.
  2. Sort every candidate into one of two lanes before scoring:
    • Incremental efficiency gain — the current process is done faster/cheaper. Cost-saving- and speed-driven (bottom-line impact).
    • Transformative innovation — a new business, product, or revenue stream that wasn't possible before current AI capabilities. Growth-driven (top-line impact). Check every candidate claimed as transformative against the Value Play taxonomy (point 3) — if it doesn't fit any of the three architectures, it's probably actually an incremental efficiency gain disguised as a big idea.
  3. For transformative candidates: check against the Value Play taxonomy. Three known architectures for creating new AI value:
    • Zero-Marginal-Cost Expertise — packaging complex specialist expertise (legal, technical, medical) into a real-time, scalable service.
    • Hyper-Personalization at Scale — the product/service becomes dynamic for every user individually (e.g. tailored learning paths, financial products).
    • Outcome-Based / Agentic Business — moving from seat-based licensing/access pricing to outcome-based pricing (e.g. billing only for a resolved ticket or a closed deal). If a candidate doesn't fit any of these and isn't clearly a combination of them, reconsider whether it belongs in the transformative lane.
  4. Score every candidate on five dimensions (1–5 per dimension, max 25 total):
    • Business Impact — measurable euro or time value (ROI, hours saved, new revenue, churn impact).
    • Technical Feasibility & AI Fit — is the problem probabilistic or deterministic in nature? Does current LLM/AI technology fit the task without unreasonable hallucination risk? (Use the SML assessment from ../task-level-decomposition-and-automation-fit/SKILL.md here if available — the problem type prediction/classification/ generation also belongs in this dimension.)
    • Data Readiness — is the needed data available, in structured form, high quality, and interfaceable? Also assess data flywheel potential: does the solution generate unique data in use that improves the model over time and reinforces competitive advantage, or is it one-off data with no self-reinforcing loop? For a deeper diagnosis (the role of data, quality/bias, validating a flywheel claim), see ../../../data-strategy-and-literacy/skills/data-role-diagnosis/SKILL.md and ../../../data-strategy-and-literacy/skills/data-ai-strategy-design-and-prioritization/SKILL.md.
    • Strategic Alignment — does the target support the organization's 1–3-year core strategy, or is it a stand-alone experiment?
    • Speed to Value & Governance/Risk — implementation time as well as regulatory risk profile (e.g. EU AI Act classification: prohibited, high risk, low risk — see ../responsible-ai-and-governance-check/SKILL.md). Also include the degree of agentic-ness here: is traditional rule-based automation enough, or does the opportunity require agentic, independent decision-making in unpredictable situations — an agentic solution is more expensive to build and govern, which slows down the Speed to Value score and should show up in it.
  5. Place every candidate on a 2x2 prioritization matrix (vertical axis: Business Impact, horizontal axis: Technical Feasibility — use the point-4 scores):
    • Quick Wins (high impact, high feasibility) — low cost, fast implementation. Active piloting candidates.
    • Strategic Bets (high impact, low feasibility) — often transformative, require significant data/architecture investment before they're worth starting.
    • Hard / Low Value (low impact, low feasibility) — high technical bar, small ROI. Avoid.
    • Deprioritize (low impact, high feasibility) — easy to do but not worth it; low value doesn't justify the resources even when implementation would be easy.
  6. Also classify the selected Quick Wins and Strategic Bets items using BCG's Deploy-Reshape-Invent taxonomy — this is a DIFFERENT question from the point-5 matrix: the matrix answers "is this worth doing and is it easy," Deploy-Reshape-Invent answers "what kind of change does this require of the organization":
    • Deploy — rolling out ready-made AI tools (e.g. copilots) for point tasks. Doesn't require process redesign.
    • Reshape — redesigning core functions and end-to-end processes around AI. Requires process change.
    • Invent — creating entirely new business models, products, and revenue streams. Requires building new business. Don't confuse this with ../ai-capability-roadmap/SKILL.md's Horizon 1/2/3 breakdown — Deploy-Reshape-Invent describes THE NATURE OF THE CHANGE (how deeply it touches the organization), Horizon 1/2/3 describes THE TIMELINE (when it's done). The same Reshape-level opportunity can land in any horizon depending on resources and dependencies.
  7. Produce the final output: a prioritized AI Opportunity Portfolio / Backlog — for every selected item: name, lane (incremental/ transformative; if transformative, which Value Play), 5D scores and total score, 2x2 position, Deploy/Reshape/Invent class. Move the 3–5 highest-priority items into ../../../business-case-and-analysis/skills/business-case-builder/SKILL.md for a deeper business case.
  8. Validate the result with stakeholders or your own experience-based checklist. Make sure in particular that opportunities aren't assessed as an isolated silo but in relation to the organization's existing strategic goals.

What this skill does NOT do

  • Doesn't make the final decision for you — it produces a structured draft to support a human decision.
  • Doesn't confirm figures, market data, or competitor data from memory — it uses the inputs you provide, or marks an assumption clearly ([assumption — verify]).
  • Doesn't assess technical feasibility in depth — the Technical Feasibility dimension here is a rough 1–5 rating, not technical due diligence. For a deeper assessment, see ../ai-use-case-feasibility-and-poc-scoping/SKILL.md.
  • Doesn't do the task-level decomposition itself — if the raw list hasn't been assembled at the task level yet, use ../task-level-decomposition-and-automation-fit/SKILL.md first.
  • Doesn't replace ../ai-capability-roadmap/SKILL.md for scheduling — it produces a prioritized list, not a scheduled roadmap.

Refinement notes

Areas to keep deepening with real practice:

  • your own rules of thumb and heuristics for this technique — e.g. which dimensions carry the most practical weight in different industries
  • concrete templates (into ../../references/, e.g. a 5D scoring table template)
  • reference cases / your own examples
  • what this skill deliberately does not do (guardrails, common mistakes) — add to the list above

Once this section is filled in and validated in practice, update the maturity field in skills_index.json to draft, validated, or canonical (see ../../../meta/maturity_levels.md). Don't add new fields to the frontmatter — name and description are the only ones allowed (see ../../../meta/frontmatter_schema.md).

Continue from here

  • Preceding skill in this pack (if a raw list doesn't exist yet): ../task-level-decomposition-and-automation-fit/SKILL.md (bottom-up) or ../ai-capability-pattern-matching/SKILL.md (top-down)
  • Next in this pack (business model design): ../ai-native-business-model-canvas/SKILL.md — designs the transition from an AI-enhanced business to an AI-native business model using an extended Business Model Canvas.
  • Next in this pack (technical validation): ../ai-use-case-feasibility-and-poc-scoping/SKILL.md — determines the technical boundary conditions of an AI use case and scopes the PoC phase.
  • Next in this pack (scheduling): ../ai-capability-roadmap/SKILL.md — places the selected items on a Horizon 1/2/3 timeline (a different question from this skill's Deploy/Reshape/Invent classification, see point 6).
  • Related skill in another pack: ../../../opportunity-recognition/skills/opportunity-value-assessment/SKILL.md — a more general, non-AI-specific opportunity assessment model.
  • If the whole process is run as a paid consulting engagement: ../ai-discovery-engagement-design/SKILL.md
  • If the client is a public-sector or non-profit body: pre-screen with ../../../specialisation-packs/public-sector-ai-service-design/skills/ps-ai-opportunity-screening-for-public-value/SKILL.md before or alongside this skill — public-value fit and mandate alignment change how the Business Impact dimension should be weighted.
  • A ready-made skill chain for this situation: see ../../../playbooks/
  • This pack's shared guardrails: ../../CLAUDE.md

References

  • ../../references/ — the pack's shared background material
  • ../../CLAUDE.md — the pack's shared guardrails