Mortgage Knowledge Base RAG#

Answer any mortgage or home-buying question instantly from a curated expert knowledge base. No hallucination — every answer is grounded in the indexed documents.

What It Does#

Given a plain-English question, this skill:

1. Searches the mortgage and real estate knowledge base using semantic similarity
2. Retrieves the most relevant passages (with source citations)
3. Synthesizes a clear, accurate answer — citing which document it came from
4. Flags gaps — if the question isn't in the knowledge base, says so rather than guessing

Use Cases#

• MLOs answering client questions quickly without Googling
• Real estate agents prepping for buyer consultations
• First-time buyers who need straight answers about the process
• Agents building FAQ content for their websites
• Verifying what a lender told you is actually correct

Inputs#

json

{
  "query": "What credit score do I need to buy a house?",
  "category": "credit|loans|process|terminology|programs|investing",
  "detail_level": "brief|standard|detailed",
  "audience": "buyer|agent|mlo|investor"
}

• query (required): Any natural-language question about mortgages, credit, or home buying
• category (optional): Narrows search to a specific knowledge domain
• detail_level (optional, default: standard): How much detail to return
• audience (optional, default: buyer): Tailors language complexity to the reader

Outputs#

json

{
  "answer": "To qualify for a conventional loan, most lenders require a minimum 620 credit score. FHA loans accept scores as low as 580 with 3.5% down, or 500 with 10% down. However, a score above 740 gets you the best rates — typically 0.5-1.0% lower APR than a 620, which translates to $150-300/month on a $350K loan.",
  "confidence": "high",
  "sources": [
    {
      "document": "First-Time Buyer Roadmap",
      "section": "Phase 1: Financial Snapshot",
      "relevance": 0.94
    },
    {
      "document": "90-Day Credit Repair Playbook",
      "section": "Home Price Math: Real Dollar Impact by Credit Tier",
      "relevance": 0.87
    }
  ],
  "related_questions": [
    "How do I improve my credit score in 90 days?",
    "What is the difference between FHA and conventional loans?",
    "How much does a 1-point credit score difference affect my mortgage rate?"
  ],
  "knowledge_gap": false
}

Knowledge Base Coverage#

The indexed knowledge base covers:

Credit & Financing

• Credit score tiers and their rate impact (real dollar math)
• The 5 credit factors and which to improve first
• FHA, conventional, VA, USDA loan programs
• Down payment assistance programs (national + state-level)
• Debt-to-income ratio calculation and limits
• What triggers hard vs. soft credit inquiries

The Buying Process

• Pre-approval vs. pre-qualification (what agents actually care about)
• Making competitive offers in different market conditions
• Inspection timelines and what to push back on
• Closing costs breakdown (who pays what)
• The 45-day closing timeline step-by-step
• What NOT to do after going under contract

Credit Repair

• Dispute letter process (FCRA rights)
• Pay-for-delete negotiation strategy
• Goodwill letter templates and when they work
• Credit utilization optimization (fastest score mover)
• Authorized user strategy
• 48-hour rapid rescore process

Market & Terminology

• 40+ glossary terms (escrow, PMI, LTV, DTI, rate lock, etc.)
• How to read a Loan Estimate (LE) vs. Closing Disclosure (CD)
• Rate lock timing strategy
• Understanding APR vs. interest rate

Lead Nurture & Follow-Up (MLO/Agent Tools)

• 90-day credit repair client journey
• Follow-up script patterns by lead stage
• Objection handling (rates, timing, affordability)

Pricing#

TierPriceQueriesSingle query$11 questionPack$710 questionsUnlimited (30 days)$19Unlimited

Technical Notes#

• Knowledge base: ~450 Q&A pairs + sourced document passages
• Retrieval: TF-IDF + semantic re-ranking
• Latency: <2s for standard queries
• Hallucination guard: returns knowledge_gap: true if confidence < 0.70
• Updates: knowledge base refreshed monthly with new loan program data