76 docs tagged with "phases"

Agents are specialized AI profiles — not generic instances. Each one has a defined role, clear limits, and a protocol for when information is insufficient.

The six operational anti-patterns of Phase 5: from launch as arrival to the frozen Context Document. With detection signals and prevention.

Six operational anti-patterns of Phase 4: the unsupervised build, optimistic parallelization, late synchronization, siloed Decision Log, track-by-track review, and the technical unanimity gate.

The five failure patterns that derail Context Phase: Over-Context, the Silent Agent, Aspirational Context, the Broken Chain, and Frozen Context.

Recurring failure patterns in Problem Phase. Knowing them significantly increases the probability of detecting them in time.

The most common errors in Solution Phase: permission disguised as consensus, solution seeking a problem, scope creep, and more.

The five Phase 5 artifacts: Signal Log, Context Update Record, Iteration Brief, KPI Dashboard, and living Context Document. With templates and creation sequence.

When the Gate Review is complete, the team earns the right to think about solutions. The Problem Statement becomes the foundation for everything that follows.

How the Solution Brief translates into AI-processable context: the transition from human thinking to Context Phase.

When the Gate Review is complete, the team has a system of Agents, Rules, and Skills. The construction phase doesn't start with a briefing — it starts with context.

Construction is not the end of thinking. It's the way to put thinking to the test. Three specific transfers connect AI Build Phase to Market Phase.

Context Debt in Phase 5 occurs when the market sends change signals and the Context Document doesn't incorporate them. It's not technical — it's semantic.

Context Drift is the gradual accumulation of technical decisions that, individually reasonable, silently move the built solution away from the original problem. In parallel construction, it distributes across tracks.

AI Build Phase effort depends on solution complexity, team structure, and construction model. Measured in Story Files completed and coordination quality, not calendar time.

Context Phase effort is measured by the quality and completeness of the Rules, Agents, and Skills system, not by calendar time invested.

Phase 5 has no duration. It has a permanent regime. What varies is the intensity of effort depending on Context Document maturity and the product's lifecycle stage.

Problem Phase effort is measured in activities completed with rigor, not in calendar weeks. Interviews, synthesis, and validation are the real units.

Solution Phase effort is measured by the depth of consensus achieved, not calendar time invested. Divergence, structuring, and validation sessions are the real units.

Context Phase is complete when the outputs of the system are recognized as correct by stakeholders — not when the documents look polished.

Problem Phase is complete when simultaneous and verifiable conditions are met — not when the team feels it understands the problem.

Conditions that must be met to pass from Phase 2 to Phase 3 Context Phase.

The AI Build Phase is complete when stakeholders recognize the entire system — not just its components — as faithful to the Problem Statement and Solution Brief.

The internal structure of the Market Phase cycle: nested loops, cadences, the signal-cause-context chain, and the Context Document update flow.

Not all market signals have the same value. The core skill of Phase 5 is reading the right data the right way and translating it into Context Document updates.

Phase 5 has no Gate Review because there is no next phase. It has a continuous health criterion indicating whether the learning system is operating correctly.

The internal structure of parallel construction: the dependency graph, the distributed Decision Log, integration synchronizations, and the QA Agent's role.

Operational guide: step-by-step process, Problem Statement anatomy, artifacts, gate review, anti-patterns, and effort estimates.

The art of understanding before solving. The phase where the team commits to understanding the real problem before thinking about solutions.

Operational guide: step-by-step process, Solution Brief anatomy, artifacts, gate review, anti-patterns, and effort estimates.

The right to build is earned by thinking. The iterative process of theorizing solutions, presenting them, incorporating feedback, and achieving real organizational consensus.

Operational guide: step-by-step process, context system anatomy, artifacts, gate review, anti-patterns, and effort estimates.

Agents, Rules and Skills: the operating system of the AI project. The discipline of translating human thinking into a system that AI can process.

Operational guide for executing Phase 4: setup, Story File-driven construction, gap management, fidelity review, and closure. With parallel construction model.

Build well because it was thought well. The phase where context materializes into a solution faithful to the defined problem.

Operational guide for executing Phase 5: measurement activation, signal capture, interpretation, Context Document updates, and iteration cycles.

The market as a permanent teacher. The phase where reality refines the Context Document and transforms hypotheses into verified knowledge.

Six artifacts that evidence faithful construction: distributed Decision Log, updated graph, synchronization minutes, Build Validation Report, Context Document, and the Iteration Brief.

Operational guides for each phase: step-by-step processes, artifact templates, gate reviews, anti-patterns, and effort estimates.

Phase 5 is progressively automated: from instrumented observation to a learning system. The precision of the Context Document determines how much automation is viable.

Rules are the norms that apply to the entire project. They are what turns a collection of independent Agents and Skills into a coherent system.

Skills are the 'what gets done concretely.' Self-contained work units with all the context an Agent needs to complete them without ambiguity.

Seven elements that produce a problem definition that is complete, actionable, and resistant to ambiguity. Includes a full example and two-level validation.

Thinking before deciding. Decomposing the problem into addressable pieces and examining the four dimensions of every AI solution.

The complete structure of the context system: project-context.md template, PRD sections, Architecture Document, Agent definitions, and Skill specifications.

Structure of the Solution Brief: the deliverable that condenses all the phase's work into a single, precise, and actionable document.

Six artifacts that form the operating system of the AI project: project-context.md, Agent Definitions, PRD, Architecture Document, Story Files, and Decision Log.

Four specific artifacts that are the evidence of rigorous work and the memory that feeds everything that follows.

Ideation Board, Solution Trees, Assumptions Register, Decision Matrix, Feedback Log, and Solution Brief.

No team enters discovery with a blank mind. These are the most common biases and the antidotes we apply to counteract them.

Construction contributes only 10% of an AI project's value. The value was already captured in previous phases. Understanding this changes how Phase 4 is managed.

A prompt is a point-in-time instruction. Context is an information architecture composed of Agents, Rules, and Skills. Confusing the two is the most common mistake.

Phase 5 is not the end of the methodology. It's the transition from a linear cycle to a learning spiral where each turn produces a more precise Context Document.

A mature Context Document — one that has survived twelve months of Market Phase — contains the deepest understanding available about the problem, the solution, and the context in which both operate.

How to manage simultaneous construction by multiple teams and AI agents without losing coherence. The dependency graph, role hierarchy, and distributed Decision Log.

Six fundamental questions structure the investigation. Each opens a different angle on the problem to compose a complete picture.

The unstoppable urge to jump to solutions is the number one enemy of good discovery. The antidote is seeking evidence that contradicts your assumptions.

The five steps of Phase 4: environment setup, Story File-driven construction, gap management, fidelity review, and Decision Log closure.

A structured process with five steps: establish Rules, design the Agent chain, generate strategic documents, decompose into Skills, and validate through outputs.

The five steps of Phase 5: measurement system activation, signal capture, interpretation, Context Document update, and Iteration Brief.

A structured process with four clear steps: kick-off, in-depth interviews, findings synthesis, and Problem Statement drafting and validation.

Five steps: divergence, structuring, convergence, external validation, and internal validation to achieve real consensus.

The team begins thinking about how AI will process the solution, without writing a single line of context.

AI Build Phase is a process of context materialization, distributed fidelity supervision, and coordinated gap management. It is not free design, nor discovery, nor unrequested improvement.

Market Phase is a continuous process of capturing, interpreting, and acting on market signals. It is not hypothesis validation, user feedback collection, or a sprint cycle.

Solution Phase is an iterative alignment process, not a hierarchical approval event.

Most failures in this phase don't come from doing it poorly, but from confusing it with something else.

The group composition changes from Phase 1: stakeholders excluded from discovery are needed for alignment.

The composition of the discovery team is a design decision that determines the quality of what you'll discover. The Actor Map changes everything.

Phase 4 introduces a functional hierarchy with five decision levels. The Context Engineer supervises, the Tech Lead coordinates, Dev Leads execute, and the QA Agent protects.

Phase 5 is not an isolated product process. It's the continuation of the organizational consensus built in Phase 2. Stakeholders from Phases 1 and 2 actively participate.

The Context Engineer is the architect of the Agents, Rules, and Skills system. Not a prompt engineer — a system designer who works across three dimensions simultaneously.

Phase 4 exists so that AI's construction speed becomes an asset, not a risk. Its function is to protect context integrity during construction.

The gap between what a team understands and what AI can process is the context. Context Phase is the discipline of crossing it.

Phase 5 exists so the market stops being a silent judge and becomes a permanent teacher. Without it, the Context Document freezes at the moment of launch.

Problem Phase exists to interrupt the pattern of building without understanding. It's the highest-return investment in the entire process.

The distance between a well-defined problem and a well-aligned solution is larger than it seems.