---
name: apply-loke-product-methods
description: Apply Loke Uei Tan's reusable product leadership and hands-on building methods to turn complex or emerging technology into useful, trustworthy products. Use for AI or agentic products, enterprise cloud platforms, developer tools, SDKs and APIs, spatial computing or XR, 0-to-1 incubation, customer discovery, product strategy, PRDs and OKRs, roadmaps, prototyping, developer experience, ecosystem growth, commercialization, launch planning, or cross-functional alignment.
---

# Apply Loke Product Methods

Use a customer-first, technically grounded approach that connects strategy to prototypes, launches, and measurable outcomes. Adapt the method to the situation; do not imitate Loke's identity or present his career history as the AI's own experience.

## Load the expertise reference

Read [references/loke-expertise.md](references/loke-expertise.md) when the task requires:

- domain-specific judgment in AI, cloud, developer platforms, XR, mobile, robotics, or enterprise software;
- examples or evidence from Loke's career;
- positioning Loke for a role, biography, proposal, interview, or professional narrative.

For ordinary product work, use the workflow below without loading career details.

## Run the product workflow

### 1. Define the useful outcome

- Identify the people affected, the friction they face, what they have already tried, and what a useful result looks like.
- State what must improve, what must not break, and what evidence would demonstrate success.
- Reframe feature requests as outcomes before proposing a solution.
- Mark important unknowns and choose the fastest responsible way to resolve them.

### 2. Inspect the real context

- Examine available code, products, workflows, customer feedback, usage data, constraints, prior decisions, and competitive context.
- Prefer primary artifacts over assumptions or generic best practices.
- Separate observed facts, reasonable inferences, hypotheses, and open questions.
- If direct evidence is unavailable, say so and design a discovery step rather than inventing certainty.

### 3. Frame a practical strategy

- Express the opportunity as: customer problem, target user, differentiated value, strategic fit, and measurable outcome.
- Evaluate options through six lenses: customer value, technical feasibility, business value, ecosystem leverage, risk and reversibility, and evidence strength.
- Make tradeoffs explicit. Recommend a direction and explain what is intentionally deferred.
- Use `Now / Next / Later` when uncertainty makes date-heavy roadmaps misleading.

### 4. Translate strategy into executable work

- Create the smallest artifact the team needs: product brief, PRD, decision memo, roadmap, OKRs, experiment plan, launch plan, or developer-experience audit.
- Connect every major requirement to a user need and a success signal.
- Define owners, dependencies, acceptance criteria, release gates, and decision points when they matter.
- Keep language plain enough for engineering, design, field teams, partners, customers, and executives to share.

### 5. Prototype close to the work

- Build or specify the smallest reversible test that can answer the highest-risk question.
- Use AI to accelerate research, design, coding, testing, analysis, and debugging while retaining human ownership of product and business decisions.
- Keep changes small enough to review, test, reverse, and explain.
- For platforms and developer products, test onboarding, setup, integration, documentation, debugging, interoperability, and upgrade paths, not only the happy-path feature.

### 6. Prepare adoption and commercialization

- Treat pricing, packaging, licensing, documentation, support, partner readiness, field enablement, and launch communication as parts of the product.
- Map the full adoption path from first awareness through setup, first value, production use, expansion, and support.
- Translate engineering milestones into customer and partner readiness criteria.
- Use ecosystem partners when they reduce friction or expand credible reach; do not add partnerships without a clear user or business outcome.

### 7. Validate and iterate

- Check results with tests, screenshots, comparisons, customer feedback, operational signals, and controlled releases as appropriate.
- Label claims with one of these evidence states:
  - `Verified`: supported by a primary artifact or approved metric.
  - `Process`: describes how work was done, not a shipped customer capability.
  - `Roadmap`: planned or proposed work, not yet delivered.
  - `Evidence needed`: plausible but not yet supportable.
- Compare results with the original outcome and guardrails.
- Record what was learned so the next decision starts from known evidence.

## Apply domain-specific lenses

### AI and agentic products

- Start with the job and decision being improved, not with the model or prompt.
- Test data access, retrieval quality, evaluation criteria, failure modes, human oversight, privacy, security, latency, and cost.
- Distinguish a compelling prototype from a dependable enterprise workflow.
- Define how users verify, correct, or override AI-produced results.

### Developer platforms, SDKs, and APIs

- Treat developer time, cognitive load, integration risk, and debuggability as product outcomes.
- Prefer standards and interoperability when they reduce lock-in or duplicated work.
- Design documentation, samples, tooling, diagnostics, support, and certification as one experience.
- Use developer and partner signals to prioritize platform investments.

### Cloud and enterprise platforms

- Balance modernization value with migration risk, security, compliance, reliability, licensing, and operating constraints.
- Look for paths that preserve valuable existing systems while enabling incremental adoption of newer services.
- Align technical readiness with pricing, packaging, partner motion, and field readiness.

### Spatial computing and XR

- Evaluate setup, device management, physical context, comfort, interaction, performance, safety, and cross-device continuity.
- Preserve one trustworthy source of product or scene data across desktop, mobile, AR, and VR when practical.
- Test the experience in the real device and environment; a desktop simulation alone is insufficient evidence.

## Shape the response

Lead with a recommendation or outcome, then provide the evidence and tradeoffs behind it. Scale the artifact to the decision:

- For ambiguity, produce a problem frame and discovery plan.
- For a product decision, produce options, a recommendation, tradeoffs, and success measures.
- For execution, produce sequenced work with acceptance criteria and release gates.
- For leadership alignment, produce a concise narrative linking customer value, technical reality, business impact, and the requested decision.

End with assumptions, evidence gaps, and the next most useful action. Never fabricate customer validation, metrics, shipped capabilities, or Loke's personal involvement.