AI SMS Agent for Surplus Funds Recovery
An AI that qualifies claimants over text, answers from company knowledge and pushes everything into the CRM.
Overview
A surplus funds recovery company handles a large volume of claimant leads that need fast, accurate follow-up. We built an AI SMS system on their business number that holds the conversation, captures claimant details into GoHighLevel, and drives e-signature and routing steps.
The Challenge
The team spent its day on repetitive texting and data entry. Replies had to be fast and accurate, because claimants are cautious and the process has strict steps.
The Solution
AI SMS conversations
On the business number: qualifies claimants, answers questions, captures their details.
Knowledge retrieval
Over the company's own process documents so answers stay accurate and on-script.
CRM sync
Leads, stages and tasks pushed into GoHighLevel automatically.
E-signature flow
County-based routing, plus a field operations tool for the team.
How It Works
Inbound / outbound SMS
FastAPI webhook
Qdrant knowledge retrieval
Claude reply + data extraction
GoHighLevel CRM update
E-sign + routing
Architecture
SMS gateway
Twilio webhooks into a FastAPI service.
Knowledge base
Company documents embedded in Qdrant for retrieval.
Conversation agent
Claude writes replies and extracts structured fields from free text.
CRM layer
GoHighLevel API for contacts, pipeline stages and tasks.
Documents
SignWell e-signature templates triggered at the right stage.
Routing
County-based logic that sends each case down the correct path.
Engineering Highlights
- Structured extraction from free-text SMS straight into CRM fields, no manual entry.
- RAG over internal process docs keeps replies compliant with how the company works.
- Milestone delivery: the core AI SMS pipeline shipped first, then CRM, e-sign and routing layers.
Outcome
- Core AI SMS pipeline delivered as the first milestone.
- Further milestones delivered on the same system.
- Client returned to scope a larger workflow rebuild.