Aakash GuptaUse these skills to supercharge your claude code setup | Oji Udezue | 3x CPO
CHAPTERS
- 0:00 – 1:34
Why mono-skilled PMs are becoming obsolete (and what replaces them)
Oji frames the core thesis: product builders can’t rely on a single skill (like “just PM” or “just code”) in an AI-accelerated world. The episode will demonstrate a future workflow where product, business, and code execution are compressed into one loop using Claude Code skills.
- •“Mono-skilled professionals” are fading as AI compresses work across disciplines
- •The modern builder needs product judgment plus technical execution
- •Claude Code can be a harness for more than just coding productivity
- •The goal is decision-quality at speed, not just faster output
- 1:34 – 2:43
What you’ll learn: layering business, product, and code skills into Claude Code
Oji explains that most public “skills repos” focus on token/cost/code tactics, but successful products require three layers: business, product, and engineering. He introduces ProductMind’s approach: product judgment and business decision skills embedded alongside code scaffolding.
- •Most AI repos optimize the coding layer only
- •Tech success requires business + product + engineering alignment
- •Skills can embed PM frameworks (problem clarity, differentiation, viability)
- •Objective: apply product judgment “on tap” while keeping code quality high
- 2:43 – 4:45
Why PMs became the bottleneck as engineering speeds up
Drawing from consulting experience, Oji argues that developers are accelerating quickly with AI while PM workflows often stay manual and slow. The mismatch turns PMs into the constraint unless they also speed up judgment, orchestration, and decision-making.
- •AI boosts early-adopting engineers faster than adjacent functions
- •PMs must accelerate orchestration and judgment to match build speed
- •Teams increasingly get pulled into “shipyard” questions: org design, workflows, new skills
- •Solution direction: reusable judgment skills and structured workflows
- 4:45 – 6:00
Live demo: scaffolding a new product idea (“Code Memo”) in Claude Code
Oji kicks off the “New Project Scaffolding” skill using a product idea aimed at helping vibe coders and teams evaluate code quality, security, and robustness. The skill runs an 11-step workflow starting with viability checks and market research before moving into repo setup.
- •Start from a business problem statement, not code
- •Skill runs a structured multi-step workflow (11 steps)
- •Early stages include viability gate and market research via web search
- •Goal: produce real artifacts (briefs, docs) before generating architecture/code
- 6:00 – 8:42
Inside the scaffolding skill: master skill + subskills (market research, tests, repo setup)
Aakash asks to inspect the scaffolding skill files. Oji shows it’s an orchestrator: a master skill plus subskills that can fetch references, run product frameworks, set up testing, and even prepare for GitHub pushes.
- •Scaffolding is a multi-skill orchestrator (master + subskills)
- •Includes market research modules and optional skill fetching
- •Bakes in PM frameworks (from Oji’s book and common PM methods)
- •Targets end state: a complete repo with docs, milestones, and engineering hygiene
- 8:42 – 11:30
The viability gate verdict and market research artifacts
The first idea passes the viability gate with moderates flagged as a de-risking agenda. The skill then produces a comprehensive market research package (competitors, pricing, trends) and a product brief with customers, value proposition, success criteria, and goals.
- •Viability gate gives a scored verdict (strong/moderate/weak)
- •Gate criteria include revenue, feasibility, differentiation, competition, user clarity, urgency
- •Market research output includes overview, trends, competitor pricing, takeaways
- •Product brief output includes problem framing, target customers, value prop, success metrics
- 11:30 – 15:34
From product thinking to code: CLAUDE.md, folder structure, testing, CI/CD, and security defaults
Oji explains what the skill produces once it reaches the engineering layer: a tailored CLAUDE.md, self-organizing repo structure, test scaffolding, CI templates, and security guardrails (like secret handling). The point is to help PMs/vibe coders hand off (or continue) with production-grade bones.
- •Generates a project-specific CLAUDE.md optimized for workflows and repo patterns
- •Creates folders for prototypes, milestones, documentation, and tests
- •Sets up continuous integration and quality systems to test on every commit
- •Adds security hygiene (e.g., preventing secrets from leaking to GitHub)
- 15:34 – 20:31
How the workflow calls sharp problem tests—and why “LLMs should say no”
Reviewing the step-by-step workflow, Oji contrasts the viability gate with deeper sharp problem tests. He emphasizes that good skills must be capable of rejecting bad ideas—something generic chat often fails to do—while still requiring human review of outputs.
- •Workflow gathers context, identifies project type, finds references, then runs gates
- •Viability gate is a compressed sharp-problem-like assessment (lane finding)
- •Sharp problem test goes deeper on value and money-making potential
- •Key principle: don’t treat outputs as gospel; review artifacts for realism and gaps
- 20:31 – 24:22
Second run: testing a weaker idea (“Standup Zero”) and when the gate warns you off
They run scaffolding on a Slack standup-digest concept and pause after the gate. The viability assessment returns weak/moderate signals: shallow problem urgency, unclear target user, heavy competition, and weak differentiation—saving time before building.
- •Skill asks clarifying questions when prompts are underspecified
- •Gate focuses attention on competition, defensibility, and urgency
- •“Strong competitive landscape” can be a negative (crowded market)
- •Outcome: proceed only with eyes open—or stop before wasting build cycles
- 24:22 – 31:28
Beyond new products: ‘Vet a Feature’ for existing product roadmaps
Aakash asks if the approach applies to features, not just new products. Oji introduces a dedicated “Vet a Feature” skill that tears apart a feature idea, identifies anti-patterns, and forces opportunity-cost thinking using sharp problem logic.
- •Most PMs work on features within existing products, not net-new apps
- •Vet a Feature evaluates whether a feature is worth building vs alternatives
- •Looks for anti-patterns and confidence gaps
- •Brings sharper prioritization discipline into AI-assisted workflows
- 31:28 – 39:32
Applying PM judgment: how to review and validate market research + PRD outputs
Oji shows how he evaluates the generated documents like a PRD review: checking competitor lists for freshness, sanity-checking positioning maps, and scrutinizing personas and scope decisions. The output is treated as a first draft requiring expert critique and refinement.
- •Credibility checks: are tools current, relevant, and correctly categorized?
- •Use differentiation maps as discussion starters, not truth
- •Review personas and segmentation (e.g., “non-technical founder” needs qualifiers)
- •Assess whether value proposition and loop match real buyer/user needs
- 39:32 – 51:59
Prototypes + customer discovery plan: turning a viable idea into testable learning
With Standup Zero deprioritized, they continue with Code Memo by generating prototypes and running a “customer discovery week” skill. The workflow explores interaction models (dashboard vs chat), then outputs a structured discovery plan with scripts, surveys, synthesis templates, and a hard requirement: recruit real target users.
- •Prototype generation to explore interaction models and the “aha” moment
- •Parallel track: customer discovery plan to de-risk unknowns
- •Discovery process: open-ended interviews → codified questions → survey for quantification
- •Explicit gate: you must have recruitable target contacts, not just founder-adjacent people
- 51:59 – 55:31
The ‘three speed problem’: code accelerates, but customer-bound work still governs truth
Oji explains that development speed is collapsing (10–20× over time), but the ‘why build’ and ‘go-to-market’ sides remain customer-bound. Skills help rebalance by compressing work into a shared repo and making product/business artifacts first-class alongside code.
- •Historically, development was the long pole; AI shortens it dramatically
- •Discovery and distribution are still constrained by real customer interaction
- •Claude feature velocity creates adoption/learning limits for users too
- •Emerging pattern: collapse Notion/Figma/GitHub separation into one shared repo context
- 55:31 – 59:20
Where to get the skills (and examples of high-leverage PM/business skills)
Oji points viewers to ProductMind’s labs and the open skills library, including scaffolding, Vet a Feature, and Vibe Memo (capturing the ‘why’ behind decisions). He highlights additional skills like roadmap-from-strategy, listening machines, scope cutting, MVP vs SLC, and pricing design.
- •skills.productmind.com for chat-access; open-source skills library for browsing
- •Vibe Memo captures decision rationale, not just code changes
- •Roadmap-from-strategy and feedback/listening skills operationalize PM craft
- •Scope cutting and MVP vs “SLC” help define the right thin slice and launch bar
- 59:20 – 1:04:41
Rolling this out in teams: shared context, central CLAUDE.md, and governed evolution
Oji warns that ungoverned AI rollouts fragment context and standards. He recommends centralizing and continuously improving a shared CLAUDE.md and harmonizing skills to produce the organization’s required artifacts—while allowing innovation to feed back into the shared system.
- •Treat CLAUDE.md like a base skill: short, essential, constantly refined
- •Avoid uncontrolled forking of skills that breaks consistency
- •Align generated artifacts to company SDLC and ways-of-working standards
- •Design rollouts so one person’s learning becomes everyone’s learning (hive-mind advantage)
