Claude Skills by Pilot2Service

Uses a ready-made, research-based library of 13 AI capability patterns (see ../../references/ai-capability-pattern-library.md) to pose diagnostic questions to a new client/industry, to assemble a raw list of AI opportunity candidates — an alternative or complement to task-level decomposition.

Builds the organization's AI capability map and roadmap from the current state to the target state across three horizons (0-6mo efficiency, 6-18mo transformation, 18-36mo new business), plus an AI Target Operating Model (ATOM) / Readiness Scorecard describing the division of labor between humans and AI.

Designs and productizes AI opportunity identification (the discovery phase) into a paid consulting engagement — a 4-phase engagement structure, fixed-price service products, and standardized deliverables (Portfolio, Business Case, ATOM/Readiness Scorecard, Roadmap).

Designs the shift from an AI-enhanced business (AI bolted onto existing operations) to an AI-native one, using an extended Business Model Canvas with four AI-specific lenses: value proposition, data moat, human-AI interaction model, and cost of compute vs. revenue. Use when deciding whether AI is an add-on to the current business model or whether the model itself needs to be rebuilt around AI.

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 / Strategic Bets / Deprioritize / Hard-Low Value) — and classifies incremental and transformative opportunities separately.

Designs a quality-control and curation process for AI-generated content or decisions — the shift from 'creator' to 'curator': what gets checked, who checks it, and against what criteria before publication or use.

Tests whether an AI opportunity is framed as automating an existing process (1st-order) or as a genuine value-chain reshuffle (3rd-order) using the shipping-container three-orders-of-effect model, before it enters scoring — catches the most common AI-strategy mistake: applying AI to unchanged structures instead of asking what AI changes about where value sits.

Scopes an AI proof-of-concept around a single falsifiable hypothesis, sets a quantitative success bar and kill criterion against a named human baseline, and builds a golden test set before development starts. Use when a PoC needs to answer a real feasibility question rather than produce an impressive demo that proves nothing about production readiness.

Structures the build/buy/partner decision for an AI capability, including hybrid options, scored on differentiation value, time to value, and total cost of ownership (inference, retraining, ML-ops talent, not just license cost). Use when an AI capability decision needs a defensible comparison instead of defaulting to \"build\" or \"buy\" by habit.

Tracks which of the organization's capabilities are becoming commoditized by AI (cheap, reproducible by a prompt) versus which are becoming newly complementary (valuable precisely because AI makes them combinable with other domains), and redirects investment away from the former toward the latter before the market forces the move.

Diagnoses whether an organization's communication structure will get mirrored in whatever AI system it builds — fragmented teams produce fragmented, siloed AI tools with no shared capability — and applies the resequencing rule: decide on org structure before building the AI architecture, not after.

Classifies an AI initiative's EU AI Act risk tier and surfaces the regulatory, data-governance, and ethics obligations that follow from it, as a first-pass structuring check rather than a legal opinion. Use before an AI initiative goes to approval, especially one touching employment, credit, biometric, or other high-risk decisions about individuals.

Identifies unauthorized/unofficial AI tool usage already happening in the organization (Shadow AI) and replaces it with a safe, scalable official solution backed by a clear ROI.

Breaks roles and processes down to task level (People Path + Process Path) and classifies every task into an Automate/Augment/Human-Only category using SML criteria (input/output clarity, cognitive nature, error tolerance, time scale) before prioritizing AI opportunities.

Converts a company's own unique expert material — a recorded workshop, a proprietary methodology session, a training deck — into an interactive AI agent customers can query, so it works as a learning aid AND a low-effort upsell/retention touchpoint instead of sitting unused as a static recording.

Extracts every load-bearing assumption behind a recommendation, traces each to its source, grades the strength of the evidence, and ranks the gaps by leverage times weakness. Use before a business case, analysis, or recommendation is locked in, to see which unverified claims could actually change the conclusion if they turned out false.

Builds a full business case — problem, options considered (including \"do nothing\"), economics (ROI/NPV/IRR), risk, stakeholders, milestone-level timeline, and recommendation — following BABOK/PMI structure. Use when a proposal needs to reach a decision-maker as a structured business case, not a pitch with only one option on the table.

Classifies requirements using BABOK's four types (business, stakeholder, functional, non-functional, transition), checks each against the IEEE 830 quality bar (correct, unambiguous, verifiable), and separates in-scope from explicitly out-of-scope items. Use before requirements are locked in, when scope keeps drifting or vague requirements like \"the system should be fast\" are passing unchallenged.

Identifies risks systematically across project phases, stakeholders, and dependencies, scores each on probability x impact, and assigns a PMI response strategy (avoid, mitigate, transfer, accept) with an owner and trigger condition. Use when a project or initiative needs a real risk register and mitigation plan instead of an informal list of worries.

Calculates ROI, NPV, IRR, and payback period against a credible do-nothing baseline, then runs a one-at-a-time sensitivity analysis to rank which two or three assumptions actually drive the result, plus upside/downside scenarios. Use when a business case's financial numbers need to survive scrutiny, not just look attractive under the base case.

Maps every stakeholder or stakeholder group on a power/interest grid, assigns RACI roles per activity with exactly one Accountable each, and cross-checks the map against the RACI for gaps. Use when an initiative needs clarity on who to engage, how intensively, and who is actually accountable for each deliverable.

Models a product or business relative to market categories: associating with an existing category, expanding/redefining a category, or assessing the creation of an entirely new category.

Maps the stages and friction points of a customer's service journey, and places AI on the journey only at the points where it genuinely creates value for the customer — not technology-first.

Structures a business, service concept, or business model into distinct functional layers (OSI-model-like) from infrastructure to brand, in order to decide which layers to build in-house and which to source through partners.

Structures competitors or alternative solutions on shared industry competitive factors, maps the industry's 'as-is curve,' and finds ways to break away from or differentiate against it — using the Blue Ocean Strategy model and the owner's 360 comparison tool.

Predicts how a specific, narrowly-defined in-group will emotionally and aesthetically react to a concept, without slow or misleading market research — by deliberately building 'exposure hours' to world-class examples, running a mental simulation before asking anyone, and validating the prediction against real feedback in a backpropagation-style loop. Use when a concept, design direction, or positioning needs a fast, defensible taste judgment before committing to build or test it.

Structures a business's activities according to Porter's value chain model into primary and support activities, in order to see where value and margin come from and where competitive advantage can be built.

Tracks promised benefits after go-live using a Benefits Dependency Network that traces each benefit back to the business and enabling changes required to deliver it, with a named owner, baseline, and target per benefit. Use once a project has moved past approval and the question shifts from \"was it approved\" to \"is the promised value actually landing\".

Turns a finished analysis into an executive-ready storyline using the Pyramid Principle: governing thought stated first, an SCQA opening, and MECE supporting arguments each traced down to specific evidence. Use when the analysis is done but still needs to become a narrative someone can act on, instead of a slide deck assembled section by section.

Builds a stakeholder communication plan by segmenting audiences on the power/interest grid and mapping each segment's position on the ADKAR change curve, defining objective, message, channel, cadence, and owner per segment. Use when a change effort needs more than a generic \"keep stakeholders informed\" line and different groups actually need different messages.

Maps the information asymmetry between a team and a decision-maker before a pitch is written: what pressures, metrics, and constraints the decision-maker is actually operating under that the team can't see, surfaced through pointed diagnostic questions rather than generic stakeholder interviews. Use before writing a pitch, business case, or executive narrative for a specific leader — not after a proposal has already been drafted around assumptions about what they care about.

Strips AI-generated corporate jargon and marketing-speak ('AI slop') out of a text or pitch and replaces it with plain, concrete, human language, using a slop-strip test, a whiteboard test (explain it as if drawing for a friend in 45 seconds), and Steve-Jobs-style one-line compression. Use as a pass over any AI-drafted or AI-assisted document, deck, or pitch before it goes to a real audience — a different register from executive-narrative-and-storyline's logical structuring.

Designs the structure of a cross-functional workshop or design-thinking session using the Double Diamond and the Diamond of Participation, with explicit divergent, groan-zone, and convergent phases and a stated decision rule. Use when planning a workshop that has to both generate ideas and actually reach a decision, not just produce \"great discussion\".

Designs a Data & AI strategy holistically (not in silos) using a Driver Tree tool to break business goals down into data points, and prioritizes what data to collect/use now vs. for the future with a Data Readiness x Strategic Value matrix. Use when an organization is planning which data and AI capabilities to invest in next.

Reads a given dataset or report critically before it's used as the basis for a decision or to train an AI model: identifies missing groups, bias types (selection, survivorship, historical, measurement, aggregation), and separates vanity metrics from decision-driving ones. Use before accepting any data-based claim, model, or recommendation at face value.

Assesses an organization's or team's data literacy with a four-quadrant framework (Understanding data, Acting on data, Engaging through data, Ethics & privacy) and a four-step maturity ladder, differentiated by role. Use before launching a data strategy or AI initiative, when you need to identify which competency gap is blocking data exploitation.

Selects and justifies a suitable data monetization model (direct: DaaS/Insight-as-a-Service/data exchange vs. indirect: product enrichment/resource optimization/risk mitigation/Data Flywheel) with a decision tree, and checks the feasibility of a Data Flywheel claim with a four-point checklist. Use once data has been validated as a strategic asset and you need to decide HOW to monetize it.

Diagnoses and justifies whether data functions in the organization as an enabler (cost, operational efficiency) or as a strategic asset (revenue-generating, monetizable, defensible) — using heuristic tests (resale, flywheel, defensibility, relevance) and the Offense/Defense framework. Use before designing a data strategy or an AI business model, when you need to determine what role data plays in the organization TODAY and what role it SHOULD play.

Translates the logic of data, an analysis, or an AI model into a story a decision-maker understands, using the Data-Information-Insight-Action ladder and the 'so what' test. Use when data or a model needs to be presented to a stakeholder so that it drives a decision instead of just sitting in a report.

Designs absolute guardrails and grounds AI outputs against a named source of truth to counter sycophancy and hallucination, weighed against \"tokonomics\" — the cost, latency, and token price of every added guardrail instruction — so safety is achieved with the minimum instruction footprint, not the maximum.

Specifies an AI's sentiment, tone, and precision (\"temperature\") as an explicit design artifact — a Behavioral Document — instead of leaving it implicit; includes how the AI should redirect rather than go silent when it hits a guardrail, and \"prompt trees\" as the dynamic replacement for a static service blueprint.

Draws an explicit boundary around decisions AI may never make autonomously (high-stakes agency), and protects the junior-expertise pipeline against being silently automated away — the apprenticeship risk that erodes an organization's future ability to supervise and validate its own AI.

Classifies an AI process against a four-level Human-in-the-Loop maturity model (Smoke-and-Mirrors, Reactive HOTL, Intentionally Designed, Collaborative Intelligence) and designs a confidence-score routing table that sends work to automation, assisted validation, or human escalation — with the override-rate red flags that catch a stalled or fake HITL setup and a named accountable calibration owner.

Audits a human-AI partnership's health with override-rate and override-accuracy metrics (catching both automation bias and an under-performing model), and designs the feedback loop that turns human corrections into structured model-improvement input — the AI Flywheel and bidirectional learning.

Protects the individual practitioner's own judgment and self-efficacy when working with an AI thinking partner: a Think-First rule (form your own hypothesis before consulting AI), an adversarial-sparring instruction pattern (ask the AI to find gaps, not agree), and voice ownership (keep authorship of your own argument). Use at the start of any analysis or working session with AI, before the first prompt is written — this is a discipline for how the human works, distinct from the rest of this ...

Maps an industry's competitive dynamics using Porter's Five Forces — rivalry, threat of new entrants, substitutes, supplier power, buyer power — to explain why the industry is structurally attractive or not, not just who the named competitors are. Use when assessing whether a market or industry segment is worth entering, defending, or exiting.

Systematically scans the market, technology, regulatory, and competitive environment across the six PESTLE categories to surface disequilibria — mismatches between what's supplied and what's genuinely wanted — before they're obvious to everyone else. Use when looking for early opportunity signals, not when validating an idea you've already committed to.

Sizes TAM, SAM, and SOM using independent top-down and bottom-up methods, narrowing each step with explicitly named constraints (geography, segment, go-to-market capacity) rather than an arbitrary percentage. Use when a market-size figure has to hold up under scrutiny in a business case or investor conversation, not just sound impressively large.

Writes the results of an opportunity assessment into a concise 1-2 page Opportunity Brief report, understandable to both the technical inventor and the decision-maker.

Restates an identified opportunity as a falsifiable claim and structurally evaluates it against fixed criteria — market viability, feasibility, resource fit, strategic fit, risk — before resources are committed. Use once an opportunity has already been noticed and needs a deliberate go/no-go judgment, not another round of scanning for more ideas.