Capabilities
What brokers and lenders actually use
Start with Criteria AI and Broker Copilot — then the applications and CDM/DSL foundation that make them reliable in production.
01 · Criteria AI
Criteria search that understands lender policy
Natural-language criteria search across lender policy — structured rules, RAG, and GraphRAG with pass/fail checks against the client profile.
- Modes for structured criteria rules, RAG, and GraphRAG
- Toggle lenders and ask about CCJs, age, LTV, or credit in plain language
- Apply the client profile for pass/fail against selected policy sources
02 · AI Broker Copilot
Broker Copilot with human–AI collaboration
Case-aware assist for status, tasks, fact-find gaps, documents, and suitability letters — with HITL queues that split work between brokers and AI agents.
- Case-aware chat for status, tasks, summaries, and suitability letters
- Fact-find and document prompts that keep journeys moving
- HITL task hubs split work between AI agents and human brokers
03 · Mortgage applications
Applications built around the model
End-to-end broker and lender apps: AIP → FMA, sourcing, KYC, submissions, and audit-ready case trails.
- Broker CRM and case workspaces wired to the CDM
- AIP → FMA promotion as a state transition, not a new record
- KYC, sourcing, and lender submission in one trail
Mortgage applications
Built into the implementation
Delivered as working software in your mortgage stack — not a standalone bolt-on.
04 · Canonical data model
One schema for every mortgage journey
A FIX-inspired mortgage CDM — universal core, product archetypes, and governed lender extensions in one schema.
- Universal Core fields present on every application
- Residential, BTL, specialist, and bridging archetypes
- Governed lender namespaces without shadowing core fields
Canonical data model
Built into the implementation
Delivered as working software in your mortgage stack — not a standalone bolt-on.
05 · DSL rule engine
Declarative rules credit teams can own
Declarative eligibility, affordability, and lender overlays evaluated against the live application object.
- Eligibility, affordability, and stage rules as first-class assets
- Lender overlays in their own namespaces
- Evaluate against the live CDM object with an audit log
DSL rule engine
Built into the implementation
Delivered as working software in your mortgage stack — not a standalone bolt-on.
Foundation
Three-layer CDM & DSL
Universal core, product archetypes, lender extensions, and a rule engine that evaluates against the live application — the spine under Criteria AI and Copilot.
RULE MaxLoanToValue · APPLIES_TO UMC · LTV MUST NOT EXCEED archetype max
Human + AI
How brokers and agents collaborate
Criteria AI and Broker Copilot accelerate policy search and case operations. HITL task queues make the split explicit — agents propose and prepare; brokers review, decide, and remain accountable.
AI agents
- Search criteria rules and policy documents
- Summarise cases and flag open tasks
- Draft suitability letters and document requests
Human brokers
- Approve or override agent suggestions
- Own advice and client recommendations
- Clear HITL tasks before submission or commitment