H·AI·K·U Glossary

Quick reference for all H·AI·K·U terminology and concepts.

B

BoltSee in paper
An iteration cycle within a unit — each bolt advances the work and produces a reviewable increment

C

Collaboration Mode
The human-AI interaction pattern: Supervised, Observed, or Autonomous

D

Deliverable
The work output of a unit — the thing being produced

E

Elaboration
The first phase: defining intent, decomposing into units, setting success criteria
Execution
The second phase: doing the work through hat-based workflows

F

Feed Forward
The mechanism by which reflection outputs inform future elaboration

H

HAIKU
Human AI Knowledge Unification — a universal framework for structured human-AI collaboration
Hat
A behavioral role assumed during execution (e.g., planner, executor, reviewer)

I

IntentSee in paper
The top-level goal or initiative — the thing being accomplished

O

Operation
The third phase: managing what was delivered
Operational Plan
An artifact produced during execution that defines how the deliverable should be managed

P

Pass
A typed iteration through Elaboration and Execution within a single intent, refining work through a specific disciplinary lens; passes are optional and profile-defined; output of one pass becomes input to the next
Phase
A stage in the lifecycle: Elaboration, Execution, Operation, or Reflection
Profile
A domain-specific implementation of HAIKU (e.g., AI-DLC for software, SWARM for marketing)

Q

Quality GateSee in paper
A configurable verification checkpoint that provides enforcement

R

Reflection
The fourth phase: learning from what happened and feeding forward

S

Success Criteria
Measurable conditions that define when work is done

U

UnitSee in paper
A discrete piece of work within an intent
Unit DAGSee in paper
The Directed Acyclic Graph formed by unit dependencies, enabling parallel execution (fan-out) and convergence (fan-in)

W

Workflow
An ordered sequence of hats that defines how a unit progresses
Workspace
A standalone knowledge base where HAIKU artifacts live — can be local, cloud-synced, or MCP-backed, and can nest hierarchically
Workspace Memory
Organizational knowledge stored in a workspace's memory directory — learnings, patterns, and domain models that compound across intents and inherit upward through workspace hierarchy