REAL ESTATE AND SALES AUTOMATION
Voice CRM Lead Management
A CRM with voice and SMS agents that work a new lead from first contact to a booked appointment, on rules the business sets itself.
- PROJECT
- Voice CRM Lead Management
- INDUSTRY
- Real Estate and Sales Automation
- STACK
- Python · Voice AI · LLM Agents · REST APIs · Webhooks · AWS · Docker
OVERVIEW
The project
A real estate sales operation was losing leads in the gap between a form submission and the first phone call. Allzone built voice and SMS agents into the CRM itself, so a new lead is called and texted on the sequence the business configures, and every conversation lands on that lead's own timeline as pipeline activity rather than in a separate chatbot log.
CHALLENGE
What made it difficult
A voice agent that books appointments is easy to demonstrate and hard to trust. It has to know the stage a lead is at, stop contacting a lead that has gone quiet, and never confirm a meeting it has not actually agreed with the person on the other end.
It also has to run the business's process rather than the vendor's. Sequence, timing and intensity of contact differ between teams, so the automation had to be configurable by the people running the pipeline instead of by an engineer.
WHAT WE DID
The work delivered.
- 01
Voice and SMS agents
Separate agents for calls and texts, configurable to run at the same time or in sequence, coordinating against one shared view of the lead's stage.
- 02
Stage aware automation
Agent behaviour defined per pipeline stage, from new lead through connected, appointment scheduled, quoted and closed, including when to hand off to staff.
- 03
Booking with real confirmation
An appointment created only once a specific time and a contact email have been confirmed by the lead, then synced to Google or Outlook with a confirmation sent out.
- 04
One timeline per lead
Every call and text timestamped on a single lead record, alongside an activity log, a generated summary and a lead score.
- 05
Operational controls
Bulk import from CSV or spreadsheet with phone and email validation, and access scoped by role across agency administrator, business staff and customer views.
- 06
Guardrails
Voicemail retries capped, contact stopping once a lead moves past the active stages, and leads with no response for thirty days dropping out of automation until reactivated.
ARCHITECTURE
How it fits together
- Conversational agents wrapped around deterministic CRM workflows, so a conversation produces a record update rather than a transcript
- Event driven integration, letting a conversation outcome trigger scheduling and downstream business actions
- Call timing, SMS timing and calendar invites all resolved in the business's own time zone
- REST APIs and webhooks connecting the agents to scheduling, qualification and downstream systems
TECHNOLOGY
The stack
- Python
- Voice AI
- LLM Agents
- REST APIs
- Webhooks
- AWS
- Docker
OUTCOME
What exists now.
A new lead is worked on the business's own sequence without a staff member dialling, and arrives as a confirmed appointment on the calendar
Sales staff read one timeline per lead covering every call and text, instead of reconstructing the history from two systems
The system stops chasing leads that have gone quiet, because the retry cap and the dormant state are part of the automation rather than an optional setting
NEXT STEP
Tell us what you're building.
Bring us the problem with its real constraints attached. We will tell you what we would build, and what we would not.