deciqAI Mortgage — Loan Program Matching
Industry front door for decision-tree. Adds domain triggers, example, packs only. Parent Process unchanged. Not lending advice. Program rules change; verify current agency/investor guidelines.
Activate when: matching a borrower to Conventional/FHA/VA/USDA/Jumbo/Non-QM; deciding fixed vs ARM; "which program fits this borrower?"; borderline DTI/LTV/credit. Do NOT activate when: borrower is pre-committed to one specific product with no alternative.
Why this variant
The parent decision-tree maps branching choices to outcomes. Program selection is a decision tree over borrower attributes (credit, DTI, LTV, down payment, property type, VA eligibility) → eligible programs → net cost/fit.
Domain inputs → parent's Process
Branch on gating attributes and roll back to best fit:
- Credit + DTI + LTV + reserves → which agencies qualify.
- Down payment / MI: FHA (MIP) vs Conventional (PMI removable) vs VA (0% + funding fee).
- Property/occupancy and loan size (conforming vs jumbo).
- Compare total cost + approval probability, not just rate.
Worked example
680 credit, 3.5% down, moderate DTI, first-time buyer. → Tree: FHA qualifies now but carries lifetime MIP; Conventional 97 may work with removable PMI. Compare 5-yr total cost + approval odds; recommend on net fit, documented. Present options fairly (ECOA).
Compliance anchors
- Present options without steering to your own benefit; fair-lending (ECOA/Reg B); ATR/QM feasibility.
Packs
- Solo LO: program-fit matrix by borrower profile.
- Brokerage: investor-overlay quick reference.
Red flags
- Matching on rate alone, ignoring MI/total cost.
- Steering toward higher-comp product (fair-lending + LO-comp risk).
- Ignoring ATR/QM feasibility.
Verification
- Eligible programs derived from borrower attributes
- Total cost + approval odds compared (not just rate)
- Options presented fairly; rationale documented
- ATR/QM feasibility checked
Part of deciqAI Knowledge Skills. Core method: decision-tree.
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/mortgage-loan-program-matching · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/mortgage-loan-program-matching.json