Diagram library
All 52 diagrams in the wiki, grouped by stage. Each one is built in code, works in light and dark themes, and has a full text description. Select a diagram to open it in context.
Start Here
- DemoToValue From demo to dependable value In: What is an AI product engineer?
- RoleMatrix Who owns which decisions on an AI feature In: AI product engineer vs AI, ML, product and full-stack roles
- Lifecycle The AI product loop In: The AI product lifecycle, from problem to iteration
- PlanTimeline The three plans on one timeline In: Choose your path: 3, 6 or 12-month plans
Product Foundations
- Matrix Where to look first In: Problem discovery for AI features
- Flow A 30-minute interview In: User interviews for engineers
- Anatomy Anatomy of a job statement In: Jobs-to-be-done for AI products
- Stack The one-page AI feature spec In: Writing a one-page AI feature spec
- Stack Four layers of metrics for an AI feature In: Defining success metrics before you build
- Decision Should this feature use a model? In: When not to use AI
LLM Capabilities for Product Decisions
- Matrix Where a model fits In: What LLMs are good and bad at
- Flow Where the time goes in a model call In: Latency budgets and perceived speed
- Anatomy From cost per call to cost per success In: Estimating cost per task
- Decision Long context or retrieval? In: Context windows, retrieval and memory trade-offs
- Flow A model choice process In: Choosing a model: a decision framework
- Decision Build, buy or customize? In: Build, buy or fine-tune
Designing AI UX
- Flow States of a streamed response In: Streaming and loading states
- Matrix Apply directly, or suggest? In: Editable outputs and undo
- Anatomy Anatomy of a cited answer In: Citations and showing sources
- Decision Answer, caveat or abstain? In: Communicating confidence and uncertainty
- Matrix How much human involvement does an action need? In: Human-in-the-loop and approval flows
- Anatomy Anatomy of a useful error message In: Error, refusal and empty states
- Flow A first-run experience that works In: Onboarding and setting expectations
Rapid Prototyping
- Flow The prompt iteration loop In: Prompt-first prototyping
- Flow A Wizard-of-Oz session In: Wizard-of-Oz tests for AI features
- Flow How a request flows through the demo In: A day-one demo in Next.js
- Decision Which output shape? In: Structured outputs and tool calls in prototypes
- Flow From prototype to spec In: Turning a prototype into a spec
Evals as Product Metrics
- Stack The eval pyramid In: Why evals are product metrics
- Flow Building a golden dataset In: Building golden datasets from real user tasks
- Anatomy Anatomy of a gradeable criterion In: Rubrics and human grading
- Flow Calibrating a model judge In: LLM-as-judge: design and calibration
- Flow Evals in the development workflow In: Regression eval suites in CI
- Flow From eval score to business outcome In: Linking eval scores to business KPIs
Shipping & Iterating
- Flow A staged rollout In: Feature flags and staged rollouts for AI
- Matrix Reading an A/B test on an AI feature In: A/B testing AI features
- Flow The life of an AI output, as events In: Product analytics events for AI features
- Stack Feedback signals, from most to least informative In: Capturing feedback: thumbs, edits and retries
- Anatomy What one trace records In: Tracing and observability for LLM calls
- Flow The iteration loop In: The iteration loop: from feedback to prompt changes
Trust, Safety & Economics
- Anatomy The three ingredients of a data-exfiltration risk In: Prompt injection and tool-use risks
- Flow Where user data goes in an AI feature In: Data privacy and retention
- Stack Layers of limits for an AI feature In: Abuse, misuse and rate limiting
- Flow Where guardrails sit In: Guardrails for inputs, outputs and actions
- Stack Monthly economics of one user In: Unit economics of AI features
- Matrix Choosing a pricing model In: Pricing AI features
- Flow From traces to cost decisions In: Cost dashboards and budgets
Projects & Career
- Anatomy Anatomy of a strong portfolio entry In: Building an AI product portfolio
- Flow The structure of a product case study In: Writing a product case study
- Flow A framework for product sense answers In: Passing AI product sense interviews
- Stack How scope grows across levels In: Career paths and leveling