Understanding Waves
Before learning the commands or the artifacts, understand WHY Waves exists and what makes it fundamentally different from every other framework.
AI Changed the Speed. Your Process Didn't.
Sprints Were Built for Human Speed
Two-week sprints made sense when a feature took a team of five developers two weeks to build. AI agents compress that same work into hours or days. The sprint becomes a container with too much empty space — ceremonies that exist because the calendar says so, not because the work demands it. Planning meetings estimate work that gets done before the meeting ends. Standups report progress on tasks completed yesterday. The framework designed to create rhythm now creates friction.
Your Agent Forgets Everything
Every AI session starts from zero. Your agent doesn't remember the architecture decisions you made yesterday, the conventions you established last week, or the bug you spent three hours debugging. You re-explain your project structure, your naming patterns, your testing approach. The agent guesses wrong, you correct it, it adjusts — and tomorrow it will guess wrong again. The cost isn't just the 30-45 minutes you waste per session. It's the compounding drift: each session, the agent invents slightly different patterns, and your codebase slowly loses coherence.
No Path From Idea to Production
You have an idea. Is it viable? How do you validate it without building it? Once validated, how do you define the product? Once defined, how do you plan the work? Once planned, how do you track implementation across sessions, across agents, across tools? Today, each of these steps uses a different tool, a different format, a different context. The AI agent that helped you analyze the market has no connection to the one writing your code. The feasibility data lives in a chat history that will be compacted. The product decisions live in your head. The knowledge chain is broken at every link.
Your Agent Treats Every Decision the Same
Renaming a variable and changing your revenue model are not the same decision. But your AI agent treats them identically — it proceeds with both, silently, without asking. It doesn't know that one is a mechanical choice and the other affects your entire business. Without a way to classify decisions by impact, you can't trust the agent with real autonomy. And without trust, every interaction becomes supervision. You're not delegating — you're babysitting a very fast typist.
What is Waves
Waves is a Product Consciousness Framework — the first methodology designed for teams where AI agents are not tools, but team members. It replaces fixed-cadence sprints with organic, variable-length delivery cycles called waves. A wave lasts as long as it needs: sometimes three days, sometimes three weeks. There are no arbitrary timeboxes, no ceremonies that exist for their own sake. Work flows at the speed the problem demands.
But Waves is not just about cadence. It is a complete product lifecycle framework that guides work from the first idea to production code through a strict artifact hierarchy. Every piece of work traces to a business decision. A product blueprint defines WHAT you're building and WHY. Roadmaps define WHEN and in what ORDER. Logbooks define HOW each piece gets implemented. Information flows downward — never duplicated upward. This hierarchy is not optional: if a logbook cannot trace to the blueprint, either the blueprint needs updating or the work should not exist.
The Golden Rule: Nothing exists in the project that is not supported in the product blueprint.
What makes Waves fundamentally different from every other framework is what happens between the artifacts and the agent. In Scrum, the framework is a social agreement: "we should update the backlog," "the PO should attend standup," "every PR should reference a ticket." Social agreements degrade. Teams stop doing retros. The backlog becomes a graveyard. Standups become status reports. In Waves, the framework is not a suggestion — it is mechanically enforced. The agent cannot skip your blueprint, cannot ignore your roadmap, cannot forget your rules. And when it encounters a decision that could affect your business, it stops and asks. This is Product Consciousness: the agent knows your product, enforces your rules, classifies its own decisions, and sees the strategic implications of every action.
The 4 Pillars
Traditional frameworks give agents instructions. Waves gives agents consciousness. These four capabilities are not features you enable — they emerge from how the framework is built.
Perception
Session Awareness
The agent starts every session knowing your product state: active wave, current phase, next objective, recent decisions, open questions. No re-briefing needed — ever. The agent doesn't need to be told to read the blueprint. It already knows.
Enforcement
Rules That Can't Be Ignored
Instructions in CLAUDE.md degrade in long sessions — it's a known LLM limitation. Waves enforces rules with real code, not suggestions. The agent can't skip your blueprint, can't ignore your roadmap, can't forget your rules. These are rails, not guidelines.
Classification
5-Level Autonomy
Every decision the agent makes is classified by impact. Technical decisions (naming, patterns) proceed autonomously. Scope changes stop and inform. Business decisions stop and project scenarios. Discoveries stop, document, and advise. The agent knows when to act and when to ask.
Metacognition
The Agent Sees What You Can't
When a primary objective completes, the agent reflects holistically before continuing. When the blueprint changes, it projects cascading impacts. When a roadmap phase completes, it produces a strategic audit. The agent watches the entire board while you focus on the immediate piece.
These are not features you configure. They are capabilities that emerge from how the framework is built. The agent doesn't choose to be conscious — the framework makes it so.
From Executor to Strategic Advisor
Waves 1.x gave structure to chaos. Waves 2.0 gives consciousness to structure.
Session Awareness
The agent starts every session knowing your product state: active wave, current phase, next objective, recent decisions, open questions. No re-briefing needed — ever.
Rules That Can't Be Ignored
CLAUDE.md instructions degrade in long sessions — it's a known LLM limitation. Waves 2.0 uses executable hooks (real code, not suggestions) that block actions when artifacts are missing. The agent can't skip your blueprint, can't ignore your roadmap, can't forget your rules.
5-Level Autonomy
Every decision the agent makes is classified by impact: Level 1-2 (technical) → proceed autonomously. Level 3 (scope change) → stop and inform. Level 4 (business impact) → stop and project scenarios. Level 5 (discovery) → stop, document, and advise.
The Agent Sees What You Can't
Three automatic triggers: when a primary objective completes, the agent reflects holistically before continuing. When the blueprint changes, it projects cascading impacts on all active work. When a roadmap phase completes, it produces a full strategic audit with recommendations.
Memory That Doesn't Fade
In long sessions, LLMs compress earlier context — losing injected rules and state. Waves 2.0 automatically re-injects your product state and classification rules after every compaction event. The rules survive the entire session, no matter how long.
Proportional to Maturity
No blueprint? The agent allows everything — you're still shaping. Blueprint but no roadmap? Blocked — create a plan first. Full artifacts? Allowed, with classification reminders on every action. Enforcement scales with your project's maturity, not with rigid rules.