New Product Development (NPD) Skill — V3

Universal Core + 10 Domain Overlays

This skill provides a two-layer framework:

• Layer 1 — Core NPD: The universal 8-stage process that applies to any product in any industry. Business + software engineering tracks at every stage.
• Layer 2 — Domain Overlay: When a specific industry is mentioned, Claude automatically loads the relevant domain reference file and applies domain-specific constraints, compliance requirements, technical patterns, and rollout considerations on top of the core framework.

How Claude Should Use This Skill

1. Detect the domain from user context (explicit mention or inference)
2. Load the domain overlay from the references/domains/ folder
3. Run the core NPD framework with domain overlay applied at each stage
4. Match technical depth to domain: hardware-heavy domains (IoT, MedTech) get firmware/edge depth; pure software domains (FinTech, EdTech) get cloud/API depth
5. Apply domain gate criteria alongside universal gate criteria

Domain detection keywords:

When domain is detected, read the relevant file in

references/domains/

before responding. When no domain is detected, use the core framework only.

NPD Process Overview (8 Stages)

Stage 1 → Idea Generation
Stage 2 → Idea Screening
Stage 3 → Concept Development & Testing
Stage 4 → Business Analysis
Stage 5 → Product Development / Prototyping
Stage 6 → Market Testing
Stage 7 → Commercialization / Launch
Stage 8 → Post-Launch Review & Iteration

At each stage, Claude applies:

• Universal activities (below)
• Domain-specific additions (from overlay file)

Stage-by-Stage Core Framework

Stage 1 — Idea Generation

Business track:

• Brainstorming: SCAMPER, How Might We, Jobs-to-be-Done
• Customer interviews and pain point mapping
• Competitive gap analysis, internal innovation sessions

Technical track:

• Engineering-led ideation: internal tools → external products
• OSS ecosystem gap analysis
• AI/ML opportunity mapping
• Platform thinking: what infrastructure can become a product?

Deliverable: Idea backlog with source, rationale, feasibility note Gate: 5–10 distinct ideas documented

Stage 2 — Idea Screening

Business track:

• Scoring matrix: strategic fit, market size, competitive advantage
• SWOT per idea; alignment with mission and resources

Technical track:

• Build vs. buy vs. integrate decision
• Stack and skills assessment
• AI/ML viability check
• Regulatory / compliance pre-check
• OSS dependency and licensing risk

Deliverable: Ranked shortlist + technical feasibility summary Gate: Top idea passes scoring matrix on all dimensions including technical risk

Stage 3 — Concept Development & Testing

Business track:

• Concept statement (problem / solution / user / benefit)
• 2–3 concept variants; concept tests; willingness-to-pay signals

Technical track:

• Time-boxed technical PoC (2–5 days) to validate core assumptions
• API / integration mapping
• LLM/AI concept validation (if applicable)
• Initial architecture sketch: monolith vs. microservices vs. serverless vs. edge
• Data availability check for AI/ML products

Deliverable: Concept brief + PoC findings + architecture sketch Gate: Users desire concept AND PoC validates technical assumptions

See

references/concept-testing.md

for validation methods See

references/technical-poc.md

for PoC patterns

Stage 4 — Business Analysis

Business track:

• TAM / SAM / SOM market sizing
• Revenue model and pricing strategy
• Cost estimation: COGS, development, marketing
• Break-even, ROI, risk assessment

Technical track:

• Cloud cost model (compute, storage, networking, APIs at scale)
• LLM cost engineering (token cost × query volume)
• Build vs. buy TCO analysis
• Security and compliance cost estimation
• Engineering team sizing by role and duration

Deliverable: Business case + cloud cost model + engineering resource plan Gate: ROI meets threshold; costs sustainable at scale; risks mitigable

See

references/business-analysis.md

for financial templates See

references/cloud-cost-modeling.md

for infrastructure cost models

Stage 5 — Product Development / Prototyping

Business track:

• MVP scope via MoSCoW prioritization
• Agile sprints; usability testing; QA feedback loops

Technical track:

Architecture & System Design:

• Cloud reference architecture selection (SaaS / AI / data platform / API / edge)
• ADRs for every major technology decision
• Database selection: OLTP / OLAP / vector / NoSQL
• Scalability: horizontal scaling, Redis caching, CDN, async queues
• API design: REST / GraphQL / gRPC + OpenAPI spec

AI / LLM Engineering (where applicable):

• LLM selection: proprietary vs. open-source; evaluation rubric
• RAG pipeline: ingestion → chunking → embedding → retrieval → reranking → response
• Prompt engineering, versioning, and A/B testing
• LLMOps: observability, fallback strategy, eval suite
• Evaluation: faithfulness, relevance, groundedness, latency, cost per query

DevSecOps:

• CI/CD pipeline (GitHub Actions / GitLab CI / Cloud Build)
• Infrastructure as Code: Terraform / Pulumi; Helm for Kubernetes
• Secrets management; SAST/DAST in CI; zero-trust IAM
• Testing pyramid: unit → integration → contract → e2e → load

Observability:

• Logs + Metrics + Traces (OpenTelemetry standard)
• SLIs / SLOs / error budgets defined before launch
• Alerting runbooks for every alert
• LLM cost dashboards and quality monitoring

Deliverable: Working MVP + ADRs + CI/CD live + observability configured + security baseline + eval suite passing + load test done Gate: Core flows work; security clean; observability live; load test passed

See

references/devSecOps.md

for patterns See domain overlay for hardware/firmware/edge specifics

Stage 6 — Market Testing

Business track:

• Beta / limited release; pricing experiments; A/B testing
• NPS, activation, retention data collection

Technical track:

• Canary deployment: 1% → 5% → 25% → 50% → 100%
• Feature flag rollout with kill-switch
• Real-load performance monitoring (p50/p95/p99 latency)
• LLM quality review on production traces
• Cost validation: actual vs. projected
• Security penetration test
• Data pipeline validation; analytics trustworthiness check
• On-call readiness verification

Deliverable: Market test report + performance benchmark + cost validation + security pen test sign-off Gate: Metrics meet targets; error rate below threshold; security passed; cost within model

Stage 7 — Commercialization / Launch

Business track:

• GTM plan: positioning, messaging, channels, pricing
• Sales and support enablement; PR and demand generation

Technical track — Launch Checklist:

• Auto-scaling configured and tested under synthetic peak load
• Rollback plan documented and executable in < 15 minutes
• Status page live
• On-call rota active with escalation paths
• Rate limiting and DDoS protection active
• All secrets rotated; IAM permissions audited
• GDPR / data privacy compliance verified
• Backups tested and recovery time confirmed
• LLM prompt versions locked; fallback model configured (if applicable)
• API reference and getting-started guide published (for API/platform products)

Deliverable: GTM plan + signed-off technical launch checklist + runbooks + docs Gate: All teams ready; technical checklist 100%; rollback tested

See

references/launch-planning.md

for GTM templates

Stage 8 — Post-Launch Review & Iteration

Business track:

• KPI tracking vs. targets; win/loss analysis; customer feedback
• Backlog prioritization for v2

Technical track:

• DORA metrics baseline: deployment frequency, lead time, MTTR, change failure rate
• LLM performance review: cost trend, eval scores, model re-evaluation
• Technical debt register: document shortcuts; prioritise paydown
• Security posture review; credential rotation; next pen test scheduled
• FinOps review: top 3 cost drivers; reserved capacity opportunities

Deliverable: Post-launch review + DORA metrics + debt register + cost plan + v2 roadmap Gate: Shipped; learnings captured; next cycle planned

Quick-Reference: Stage Summary Table

Domain Overlays Available

Load the relevant file from

references/domains/

iot-manufacturing.md

healthcare-medtech.md

fintech-banking.md

retail-ecommerce.md

agritech-food.md

edtech-learning.md

logistics-supplychain.md

energy-cleantech.md

real-estate-proptech.md

government-smartcity.md

Core Reference Files

• references/concept-testing.md

  — Stage 3 validation methods

• references/business-analysis.md

  — Financial modeling, Stage 4

• references/cloud-cost-modeling.md

  — Cloud + LLM cost engineering

• references/launch-planning.md

  — GTM and launch templates

• references/technical-poc.md

  — PoC patterns and LLM evaluation

• references/devSecOps.md

  — CI/CD, security, observability patterns