The Coach Oracle

The Problem

When an investor pays the Engagement Fee, Property Coaches at ProPhero historically needed 3–4 onboarding meetings before being able to discuss a real investment strategy. Each session required the coach to manually cross-reference HubSpot (client profile, budget, preferences) with Vistral (property catalog), build a tailored shortlist, and craft a strategy from scratch — a slow, repetitive, and error-prone process that delayed time-to-value for both client and advisor.

The Solution — Coach Oracle

Coach Oracle is an automated, AI-driven pipeline that fires the moment a HubSpot deal transitions to Engagement Fee Paid. In under 60 seconds it produces a personalized investment strategy and pre-assigns a ranked shortlist of properties to the engagement — so the coach lands in their first call fully briefed, and the client lands in the portal with curated options already waiting.

Goal: Replace 3–4 onboarding meetings with 1 strategy-driven call.

How It Works

1. Trigger — HubSpot webhook hits the prophero-customers BFF on the Engagement Fee Paid stage transition.
2. Parallel tracks — Activation flow sends a welcome email; Coach Oracle pipeline starts.
3. Data orchestration — BFF reads the client's preferences from the Engagements table (HubSpot-synced): budget range, cash available, borrowing capacity, country/area preference, investment objective, property type, condition, and financing.
4. Hard filtering — Operation Data Service narrows the catalog by budget, geography, typology, condition, and availability.
5. AI ranking with Claude — Claude receives the client profile, filtered candidates, and supplementary product catalog (Value Hero, Ireland, ValueGiro), and returns:
   • An investment strategy (headline, summary, capital allocation, leverage analysis, phasing)
   • Top 5 properties ranked with one-line rationale each
   • Transparent matching parameters
   • "McDonald's combo" logic: when budget allows, at least 2 products are proposed (property + supplementary)
6. Slack briefing — A Block Kit message lands in the coach's channel with strategy, client highlights, matching parameters, and the ranked top 5 (name, location, price, gross yield, typology, link).
7. Property assignment — assignPropertiesToEngagement Lambda writes the ranked IDs into engagements.properties so the portal can show them.
8. Client portal experience — Gated by the coach_oracle feature flag, the client skips the activation tasks page and lands directly on the PLP with pre-selected properties, a banner inviting them to schedule a call, and reservation/discard actions gated until the strategy call is completed.
9. Refresh loop — /refresh-client [deal_id] in Slack re-reads HubSpot, re-filters, re-ranks via Claude, replaces the assigned properties, and replies in the same Slack thread.

Architecture at a Glance

HubSpot (deal stage change)
  → prophero-customers BFF
    → Engagements table (HubSpot-synced)
    → Operation Data Service (property filter)
    → Claude API (strategy + top 5 ranking)
      → Slack (coach briefing)
      → assignPropertiesToEngagement → engagements.properties
        → Portal PLP (curated landing experience)

Expected Results

• −20% average sales cycle length
• +30% session → proposal conversion rate
• +25% advisor satisfaction (internal NPS)
• Onboarding compressed from 3–4 meetings to 1 strategy-driven call
• Zero manual cross-referencing between HubSpot and Vistral
• Coaches start every engagement fully briefed; clients start with a curated, personalized portal experience