Skip to content

Federation Protocol

Federation is the endgame: intelligence flows across organizations while raw data stays home.

The Vision

Traditional enterprise collaboration: 

Enterprise A wants to share insights with Enterprise B
    ↓
Option 1: Email documents (insecure, no audit)
Option 2: Data room (expensive, slow)
Option 3: API integration (tight coupling, vendor lock-in)

Archivus Federation: 

Enterprise A shares verified claims from Knowledge Graph
    ↓
Enterprise B queries claims via federation protocol
    ↓
Hedera anchors provide cryptographic proof
    ↓
No documents transferred, no data rooms, no raw data exposure

Core Principle: Data Sovereignty

┌──────────────────────────────────────┐
│         ENTERPRISE A                  │
│                                       │
│  Raw Data:          STAYS HOME        │
│  ├─ Documents       ✗ Never shared   │
│  ├─ Database        ✗ Never shared   │
│  └─ File storage    ✗ Never shared   │
│                                       │
│  Knowledge Graph:   MAY FEDERATE      │
│  ├─ Entities        ✓ Canonical IDs  │
│  ├─ Claims          ✓ With provenance│
│  ├─ Relations       ✓ Anonymized     │
│  └─ Trust scores    ✓ From GOLAG     │
│                                       │
└──────────────────────────────────────┘

The rule: Intelligence flows, data stays home.

What Federates

1. Verified Claims

{
  "claim_hash": "sha256:a91f3b4c5d6e7f8a...",
  "claim_text": "Product X has ISO 9001 certification",
  "confidence": 0.92,
  "source_type": "document",
  "verified_at": "2026-02-01T10:00:00Z",
  "hedera_anchor": {
    "topic_id": "0.0.12345",
    "tx_id": "0.0.12345@1234567890.123456789",
    "consensus_time": "2026-02-01T10:00:05Z"
  },
  "about_entities": ["ent:product:x:uuid"]
}

What's NOT included: Source document, internal notes, raw extracted text.

2. Entity References

{
  "entity_id": "ent:company:acme:uuid",
  "canonical_id": "wikidata:Q12345",
  "entity_type": "organization",
  "name": "Acme Corporation",
  "aliases": ["Acme Corp", "Acme Inc"],
  "wikidata_qid": "Q12345"
}

Key feature: Canonical IDs (Wikidata QIDs, DUNS numbers) enable cross-organizational entity linking without exposing internal data.

3. Relationship Signals

{
  "from_entity": "ent:person:john:uuid",
  "to_entity": "ent:company:acme:uuid",
  "relation_type": "employed_by",
  "confidence": 0.89,
  "temporal_validity": {
    "valid_from": "2020-01-01",
    "valid_until": null
  }
}

Anonymization: Can be shared as "Person with role X employed by Company Y" without revealing identity.

4. Trust Scores

{
  "claim_hash": "sha256:a91...",
  "golag_verification": {
    "verified_by_agents": 5,
    "consensus_confidence": 0.94,
    "expert_agents_voted": 3,
    "contradictions_resolved": 1
  }
}

Value: Federated organizations see calibrated confidence from evolutionary agent verification.

Federation Handshake

Phase 1: Discovery

Enterprise B discovers Enterprise A offers federated claims
    ↓
Request: GET /api/federation/capabilities
    ↓
Response:
{
  "federation_version": "1.0",
  "offers": ["verified_claims", "entity_refs", "relation_signals"],
  "domains": ["manufacturing", "quality_assurance"],
  "trust_level": "limited" | "full",
  "hedera_topic": "0.0.12345"
}

Phase 2: Trust Establishment

Enterprise B requests federation access
    ↓
Enterprise A reviews:
  - Organization identity
  - Requested scope (what claims?)
  - Duration (temporary or permanent?)
    ↓
Enterprise A approves with trust level:
  - "limited": Read verified claims only
  - "full": Read + contribute claims

All trust relationships logged for audit.

Phase 3: Query Protocol

Enterprise B queries Enterprise A's federated claims
    ↓
POST /api/federation/query
{
  "scope": "claims",
  "filters": {
    "entity_types": ["product"],
    "confidence_min": 0.80,
    "about_entities": ["wikidata:Q12345"]
  }
}
    ↓
Enterprise A returns matching claims with Hedera anchors
    ↓
Enterprise B verifies via Hedera (trustless)

Trust Chain Recording

Every federated interaction is logged:

{
  "event_type": "claim_query",
  "requesting_tenant": "tenant_b_uuid",
  "providing_tenant": "tenant_a_uuid",
  "scope": "verified_claims",
  "filters_applied": {...},
  "claims_returned": 47,
  "timestamp": "2026-02-07T14:30:00Z",
  "trust_level": "limited",
  "hedera_anchor": "0.0.12345@1234567890.123456789"
}

Audit requirement: Both organizations can prove what was shared and when.

Use Cases

1. Supply Chain Intelligence

graph LR
    MFG[Manufacturer] -->|Verified Claims| DIST[Distributor]
    DIST -->|Aggregated Intelligence| RETAIL[Retailer]

    MFG -.Documents stay home.-> MFG
    DIST -.Documents stay home.-> DIST
    RETAIL -.Documents stay home.-> RETAIL

What flows: - Product demand forecasts (verified claims) - Quality certifications (entity properties) - Supplier relationships (relation signals)

What stays home: - Internal sales data - Pricing agreements - Strategic plans

2. M&A Due Diligence

Acquiring Company needs to verify target's claims
    ↓
Target Company grants temporary federation access (90 days)
    ↓
Scope: Contracts, liabilities, material agreements
    ↓
Acquiring Company queries Knowledge Graph:
  - "Material contracts expiring within 12 months?"
  - "Outstanding liabilities?"
  - "Key customer dependencies?"
    ↓
Results: Verified claims with Hedera anchors
    ↓
No data room needed, no document transfer, full audit trail

┌─────────────────────────────────────┐
│          LAW FIRM                    │
│                                      │
│  Manages 100+ client Knowledge Graphs│
│  (each client's data isolated)       │
└─────────────────────────────────────┘
          │
          │ Federation Query
          ▼
┌─────────────────────────────────────┐
│  "Find change-of-control clauses     │
│   across all client contracts"       │
└─────────────────────────────────────┘
          │
          ▼
┌─────────────────────────────────────┐
│  Results:                            │
│  ├─ Client A: 23 contracts           │
│  ├─ Client B: 12 contracts           │
│  └─ Client C: 8 contracts            │
│                                      │
│  No raw documents transferred        │
│  Each client's data sovereignty      │
│  maintained                          │
└─────────────────────────────────────┘

4. Industry Consortiums

Example: Manufacturing quality data sharing

10 manufacturers agree to share quality defect data
    ↓
Each maintains sovereign Knowledge Graph
    ↓
Consortium queries:
  - "Defect patterns for component X?"
  - "Supplier quality trends?"
  - "Industry-wide failure modes?"
    ↓
Aggregated insights WITHOUT exposing:
  - Individual company volumes
  - Specific customer names
  - Internal processes

Value: Network-level insights while preserving competitive confidentiality.

Security Model

Access Control

Federation permissions:
├─ Scope: What can be queried (claims, entities, relations)
├─ Filters: What subset is accessible
├─ Duration: How long access lasts
├─ Trust Level: Read-only vs read-write
└─ Audit: All queries logged

Principle: Explicit grants, no implicit access.

Data Minimization

Only return what's necessary:

Bad: 

{
  "claim": "Contract value $1.5M",
  "internal_notes": "CEO approved after negotiation",
  "source_document": "contract_final_v7.pdf"
}

Good: 

{
  "claim_hash": "sha256:a91...",
  "claim_text": "Contract value $1.5M",
  "confidence": 0.95,
  "hedera_anchor": "0.0.12345@1234567890.123456789"
}

Anonymization

When sharing relationship signals:

Direct (requires explicit permission): 

"John Smith employed by Acme Corp"

Anonymized (default): 

"Person with role 'Senior Engineer' employed by Organization in 'Technology' sector"

Hedera Verification

The Trust Mechanism:

Enterprise A shares claim
    ↓
Enterprise B receives claim + Hedera anchor
    ↓
Enterprise B queries Hedera mirror node:
  GET https://mainnet-public.mirrornode.hedera.com/api/v1/topics/0.0.12345/messages/123
    ↓
Response includes:
  - Consensus timestamp
  - Message content (Merkle root or claim hash)
  - Sequence number
    ↓
Enterprise B verifies:
  1. Claim hash matches anchor
  2. Merkle proof valid (if batched)
  3. Timestamp is authentic
    ↓
Result: Trust WITHOUT trusting Enterprise A

Breakthrough: Cryptographic proof replaces institutional agreements.

Federation Tiers

Tier Federation Access
Free/Starter/Pro None
Team Read-only (view federated claims from partners)
Enterprise Full (read + write, create consortiums)

Network Effects

As more organizations federate:

1-10 Organizations: - Pairwise intelligence sharing - Manual trust establishment

10-100 Organizations: - Industry consortiums emerge - Shared canonical entity IDs - Network-level insights

100-1000 Organizations: - Knowledge marketplace - Verified fact feeds (subscriptions) - Cross-industry intelligence

1000+ Organizations: - Global knowledge graph - AI trained on verified facts - The TCP/IP of enterprise intelligence

What This Enables

Traditional Collaboration Federated Intelligence
Share documents via email Share verified claims via graph
Data rooms for due diligence Federated queries with Hedera proof
API integrations (tight coupling) Protocol-based federation (loose coupling)
Trust institutional agreements Trust cryptographic proof
Vendor lock-in Vendor-independent verification
Siloed intelligence Network-level insights

The Endgame

A world where: - Organizations share what they know without sharing what they have - Cryptographic proof replaces institutional trust - Industry consortiums aggregate intelligence at scale - Cross-organizational AI is verifiable, not just fluent

Not "trust us"—verify independently via Hedera.

Intelligence federates. Data stays sovereign.