End-to-End Intelligent Workflow — [Client Brokerage]

AI augments the advisers — it does not replace the advisory work. The goal: cut the 60–70% admin load so brokers spend more time advising clients and winning business.

~150 enquiries / month
~30 applications / month
6 brokers
60–70% of case time on admin
60–70% of active-case time is administration, document handling, lender comms & CRM updates. Only 30–40% is client advice. Most delay is waiting for information, not mortgage advice.
1

Lead Generation & Capture

2

Initial Contact / Fact Find

3

Research & Recommendation

4

Document Collection

Primary Bottleneck
5

Application, Submission & Underwriting

Client

Enquiries via website forms, live chat, phone, email, referrals, guides & campaigns

Contacted by broker; provides requirements & financial details

Sends documents: identity, income, banking, property

Specialist cases require 30–50 separate items. Documents arrive via email attachments, phone photos, PDF scans — no single client portal today.

Responds to queries; kept updated during underwriting

AI Platform
Phase 2

Auto-capture & CRM enrichment from phone/email leads

Phase 2

AI-assisted fact find capture; auto CRM update after calls/meetings

Phase 2

Lender matching & case summary generation

Phase 1 — Build Now

Intelligent Document Engine

  • Secure upload & collection
  • Completeness & quality check
  • OCR & field extraction
  • Automated chasing (email / SMS / WhatsApp)
  • Auto CRM updates
Phase 1

Stalled-case monitoring & missing-action alerts

Phase 2

Query drafting, compliance notes & client updates

Broker

Lead assigned by admin team

Full fact find · Affordability assessment · Explain costs & process

Research lenders · Discuss unusual criteria internally · Source products · Prepare recommendation

Sends terms of business + document request list

Complete application · Upload docs to lender portal · Answer queries · Coordinate valuation

Admin / Client Services

Qualify leads & allocate to brokers

Chases clients for outstanding documents

HubSpot CRM

Lead record created

Auto: web forms / chat · Manual: phone / email

Fact find data recorded

Document fields populated via OCR extraction

Case status tracking through to completion

Lender / Bank

Submission → Underwriting → Offer → Completion

Each lender has its own portal and process. No universal lender API — the platform augments, it does not replace lender portals.

Workflow Notes

  1. Pain · Stage 1 Phone & email leads added to HubSpot manually.
  2. AI · Phase 2 · Stage 1 Auto-capture & CRM enrichment — phone and email leads automatically captured and enriched in CRM.
  3. Pain · Stage 2 Multiple attempts to reach clients; many are just information-gathering, not ready to move — high admin cost.
  4. AI · Phase 2 · Stage 2 AI-assisted fact find capture; auto CRM update after calls/meetings.
  5. AI · Phase 2 · Stage 3 Lender matching — suggests suitable lenders from case characteristics & criteria.
  6. AI · Phase 2 · Stage 3 Lender-ready case summaries — auto-generated submission packs.
  7. AI · Phase 1 · Stage 4 Intelligent document collection — one simple upload point; client (or their assistant) drops files by portal, email or WhatsApp.
  8. AI · Phase 1 · Stage 4 Automatic completeness & quality check — instantly flags wrong doc, missing pages, poor photo before it reaches a human.
  9. AI · Phase 1 · Stage 4 OCR & extraction — key fields read from documents and written straight into HubSpot.
  10. AI · Phase 1 · Stage 4 AI chasing — polite, automated reminders for outstanding items via email / SMS / WhatsApp until the file is complete.
  11. AI · Phase 1 · Stage 4 Auto CRM updates — HubSpot kept current from emails, calls & document activity.
  12. Pain · Stage 4 SINGLE LARGEST DELAY. Clients send docs over several weeks, wrong versions, low-quality photos, missing pages, forgotten items — endless follow-up calls & emails.
  13. AI · Phase 1 · Stage 5 Stalled-case monitoring — proactively highlights cases going quiet or missing an action.
  14. AI · Phase 2 · Stage 5 Query drafting — draft responses to common client & lender questions.
  15. AI · Phase 2 · Stage 5 Compliance notes & audit trail — generated automatically from broker interactions.
  16. Pain · Stage 5 Fragmented lender portals; brokers work across many different systems.