An AI phone concierge that turns inbound calls into CRM leads
A voice agent that answers inbound seller calls, qualifies them on a fixed template and writes the lead straight into the CRM.

- Client
- A US voice-AI agency
- Year
- 2026
- Status
- Delivered
- Timeline
- One-week proof of concept
The problem
The agency’s customer takes inbound calls from property sellers. Calls were handled by hand, and qualifying each caller and logging the lead took a person every time.
What we built
- A call pipeline: call source, voice agent, extraction of the call result, then the CRM record
- A bilingual voice agent on Retell with a versioned playbook and a Hot, Warm or Cold rubric
- An operations app with calls, the agent’s settings and draft, try and publish versions you can edit in the app
- Call screening with an allowlist, business hours and a live hand-off to a person
- Invite-only email-code sign-in, a Salesforce connection and a stand-in CRM for testing
- Test calls from a real phone line or straight from the browser
Stack
- Retell
- Next.js
- Postgres
- Resend
- Salesforce
- Vercel
Where it landed
- Built against a written acceptance test: inbound call, AI conversation, populated CRM record
- 23 recorded design decisions, so the next builder can see why each choice was made
Want something like this?
Tell us the idea in a few lines. We will say what week one looks like.


