Unclassified

DOCTRINE

INTEGRATEWISE — COMPLETE PRODUCT DOCTRINE
Build Brief for Lovable
Version: Final — May 2026

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 1 — THE CATEGORY
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE COMPANY DESCRIPTION

One line:
IntegrateWise is the continuity layer for the AI era.

Long form:
IntegrateWise is the continuity layer for the AI era —
connecting systems through a shared protocol, preserving
organizational memory in a unified Spine, and governing how
intelligence reasons and acts across tools, models, and time.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 2 — THE ROOT CAUSE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Modern software was built for transactions.
Organizations operate on continuity.

Transactions record what happened.
Continuity preserves why it happened, what it means,
and how it should influence every future decision.

As organizations adopt more tools and more AI,
continuity breaks:

Context fragments.
Memory disappears.
Knowledge decays.
Decisions lose lineage.
Humans become translators between systems.

IntegrateWise exists to restore continuity.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 3 — THE 18 PROBLEMS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Every problem below is a continuity failure.

Human is the integration layer
→ Operational continuity fails
NxN connection explosion
→ Protocol continuity fails
AI starts cold every session
→ Memory continuity fails
No persistent organizational memory
→ Knowledge continuity fails
Stale data across tools
→ Data continuity fails
Unstructured data has no intelligence
→ Semantic continuity fails
Writes are ungoverned
→ Governance continuity fails
Humans don't know what they don't know
→ Awareness continuity fails
AI executes without human approval
→ Decision continuity fails
System degrades over time
→ System continuity fails
Context lost when models change
→ Model continuity fails
No single operational surface
→ Operational continuity fails
Continuity breaks when things change
→ All continuity contracts broken simultaneously
Knowledge stays static while world evolves
→ Learning continuity fails
Cold start kills adoption
→ Value continuity fails
Best people carry the most institutional risk
→ Relationship continuity fails
Tool lock-in holds data hostage
→ Tool continuity fails
AI wastes tokens on code instead of reasoning
→ Intelligence continuity fails

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 4 — THE ARCHITECTURE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Three layers. Each necessary. None sufficient alone.

LAYER 1 — PROTOCOL (MCP)
Purpose: How everything connects.
Solves: Discovery, invocation, standardization,
agent-to-agent communication, ecosystem reach.
MCP is the language.
Gets stronger as MCP ecosystem adoption increases.

LAYER 2 — MEMORY (Spine)
Purpose: What everything remembers.
Solves: Context, continuity, organizational memory,
entity identity, knowledge persistence.
Spine is the memory.
Model-independent. Survives tool changes.
Survives provider changes. Survives team changes.

LAYER 3 — INTELLIGENCE (Agents)
Purpose: What everything understands.
Agents: Twin · Signal Engine · Triage Bot · Operator · Maintainer
Solves: Reasoning, governance, monitoring,
proposals, execution, preservation.
No Python. No DataWeave. No generated transformation code.
LLM tokens are reserved for reasoning only.
Deterministic work is handled deterministically.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 5 — THE CONTINUITY MAP
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Every component has one job: preserve a type of continuity.

MCP → continuity of communication
Spine → continuity of memory
Entity Graph → continuity of relationships
Lineage → continuity of decisions
Signal Engine → continuity of awareness
Twin → continuity of human understanding
Triage Bot → continuity of governance
Operator → continuity of execution
Maintainer → continuity of system health
Knowledge Evol. → continuity of learning

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 6 — THE THREE ENGINES
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE FACE: Twin (conversation, presentation, action)
THE BRAIN: Three Engines (classification, governance, monitoring)
THE BODY: Spine (structure, memory, relationships)

Engine 1 — The Normalizer
WHEN: On arrival — every piece of data that enters
WHAT: Classifies · Links · Extracts · Files · Triggers
DOCTRINE: Semantic Contract (Law 2)
All data passes through. No exceptions.

Engine 2 — The Triage Bot
WHEN: On every write — nothing enters Spine without passing through
WHAT: Deduplicates · Conflict-checks · Scores · Routes · Records lineage
ROUTING:
HIGH confidence → auto-approve → write to Spine
MEDIUM → route to human review
LOW → reject or ask for clarification
DOCTRINE: Governance (Law 5) + Lineage (Law 7)
Human reviews ~20% of writes. 80% flows automatically.

Engine 3 — The Signal Engine
WHEN: Continuously — always watching
WHAT: Monitors every entity · Detects deviation ·
Classifies urgency · Surfaces to Twin
DOCTRINE: Pattern Detection + Drift Detection
Not reactive. Proactive. Always on.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 7 — THE TWO INGESTION PATHS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Path 1 — MCP Path (speaks MCP natively)
App → MCP Gateway → Validation → Resolver → Spine

Path 2 — Connector Path (API wrapper)
App → Connector → Loader → Normalizer → Resolver → Spine

Both paths converge at the Resolver.
Both paths lead to one destination: the Spine.
The path an app takes does not matter to the rest of the system.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 8 — THE SPINE (TWO LAYERS)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

CLOUDFLARE (external interface — what the world sees)
├── D1 hot cache, governance, fast queries
├── KV entity cache, TTL 60-300s
├── R2 blobs, artifacts, exports
├── Workers compute, protocol, routing
└── Queues async processing

SPINE DB (internal canonical store — never exposed)
├── entities
├── org_memory
├── conversational_memory
├── pgvector embeddings
└── accessed only through Cloudflare Workers

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 9 — THE SEVEN LAWS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

LAW 0 — PRIMACY
The human is the authority. The system serves human intent.
Never replaces human judgment.

LAW 1 — PROTOCOL
Every agent connects through the Protocol Gateway.
No agent connects directly to an external tool.

LAW 2 — SEMANTIC
Every piece of data that enters the Spine passes
through the Normalizer. No raw, unnormalized writes.

LAW 3 — CONTINUITY
Every agent reads from the Spine first.
No agent reads directly from an external tool
when the Spine has the data.

LAW 4 — CAPABILITY
Every agent thinks in capabilities, not tools.
Create a ticket. Send a message. Update a record.
The tools change. The capabilities stay.

LAW 5 — GOVERNANCE
AI proposes. Human approves. Operator executes.
Nothing runs without human approval.

LAW 6 — MEMORY
The human's context is permanent.
Not dependent on model, provider, or session.

LAW 7 — LINEAGE
Every action, every decision, every execution is traceable.
Accountability serves authority.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 10 — THE STRATEGIC PROPERTIES
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Deterministic work is handled deterministically.
LLM tokens are reserved for reasoning only.
No Python. No DataWeave. No generated transformation code.
The Spine owns truth. Tools are interchangeable.
Human approval is required before anything executes.
The architecture gets stronger as MCP adoption increases.
(Competitors are threatened by standardization.
IntegrateWise benefits from it.)
Knowledge survives tool changes, model changes, team changes.
The system compounds — every cycle makes it smarter.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 11 — THE THESIS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

"Without MCP, the Spine is isolated.
Without the Spine, MCP is just a better REST API.
Without Intelligence, both layers sit inert.
Together — this is the network layer for organizational context."

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 12 — THE OUTCOME
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

The organization becomes independent of:

Which tools it uses today
Which AI provider it chose last year
Which employees carry institutional memory
Which workflows were built by people who have left

Knowledge survives.
Context survives.
Decisions survive.
Relationships survive.
The organization survives change.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
END OF DOCTRINE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━