The problem
Our client acquires and operates residential communities, raising capital from private investors on one side and buying properties directly from owners on the other. Both sides ran on the same CRM, with long email and SMS nurture campaigns already in place. We were brought in because those campaigns generated replies that nobody had time to answer.
Inbound replies from investors and property owners sat unanswered until the founder personally got to them, often days later.
The founder was the only person who could answer investor questions, making him the bottleneck on every warm lead.
Nurture sequences kept sending scheduled marketing emails to people who had already replied, because no reply detection existed.
Interested leads had to be qualified manually before a meeting, so unqualified prospects still consumed calendar time.
There was no alert when a lead asked to speak to a human, so hot leads could go cold in the inbox.
What we built
We built two separate AI sales agents rather than one, because an investor conversation and a property-owner conversation need different knowledge, different tone and very different compliance rules. Both agents were written against the client's own brand voice documentation so replies read like the founder rather than a chatbot. Because the offering is a regulated securities product, we treated compliance as a hard constraint in the agent instructions rather than something to review after the fact.
Investor enquiry
Email thread
Lead
AI agent
Lead
What the system did
- Contact removed from the nurture sequence
- Eligibility form sent
- Meeting booked with the founder
Investor AI agent
Answers investor questions from a curated knowledge base, refuses anything outside it, and never projects or guarantees returns.
Seller AI agent
Holds a low-pressure conversation with property owners, captures a phone number, and escalates any genuine interest to the founder.
Qualification form
Confirms investor eligibility before booking, routing qualified prospects to the calendar and declining others politely.
Booking flow
Qualified prospects move straight from the form into the founder's calendar without human involvement.
Escalation alerts
When a lead asks for a person, the conversation is assigned to the founder and an in-app plus SMS alert is sent immediately.
Opt-out handling
An opt-out stops the agent and suppresses all outbound channels for that contact automatically.
How it works
From trigger to result.
How the system works
7 stages
01Lead enters CRM
TriggerFrom a campaign, form or event list, tagged investor or seller.
02Agent assignment
LogicThe tag assigns the matching AI agent.
03Reply handling
AIReads the thread, retrieves from the knowledge base, answers in the firm's voice.
04Nurture exit
LogicThe reply removes the contact from scheduled marketing.
05Qualification form
LogicEligible prospects continue. Others receive a courteous decline.
06Calendar booking
ResultThe booking writes back to the CRM pipeline.
07Human escalation
PersonA request for a person triggers an in-app and SMS alert to the founder.
Our approach
How we worked.
Discovery
Audited the existing CRM, campaigns and pipelines to find what already worked and what was missing.
Seller build
Built and tested the seller agent, escalation alerts and pipeline automation first, as the lower-risk side.
Investor build
Built the investor knowledge base, compliance rules, qualification gate and booking chain.
Testing
Ran scripted scenarios against the live system using internal accounts only, never real leads.
Handover
Documented the build and agreed the next phase with the client.
The outcome
What changed.
Inbound replies now get an accurate answer within minutes instead of days, including outside business hours.
The agent refuses to guess: when a question falls outside the knowledge base it says so and escalates rather than inventing an answer.
Eligibility is confirmed before a meeting is booked, so the founder's calendar only fills with prospects who qualify.
The founder is alerted the moment a lead asks to speak to a person, instead of finding out later in the inbox.
Questions
What people ask.
Can an AI agent be used in regulated investor conversations?
It can support them when the limits are written into the agent and a person stays responsible. We instruct the agent never to promise or project returns and never to give investment, legal or tax advice. Anything outside those limits is handed to a person. The firm's own compliance review still applies to everything the agent is allowed to say.
Will the AI sound like our firm or like a bot?
We write the agent against your own brand voice material and your existing emails, not a generic template. We then test it on real conversations and correct anything that reads off-brand before it ever speaks to a lead.
What happens when the AI does not know the answer?
It says so plainly and hands the conversation to a person. We deliberately build agents that refuse to guess, because in regulated or high-value sales a confident wrong answer costs far more than a short delay.
How do you stop the AI from contradicting our existing email campaigns?
We connect the agent to your nurture sequences so that a reply removes the contact from scheduled sends. Without that, a lead can be holding a live conversation with the agent while still receiving automated marketing emails.
Do we need to replace our CRM to do this?
No. We build inside the CRM you already run, and we add new automations alongside your existing ones rather than rewriting them. That keeps live campaigns and real contacts untouched while the new system is tested.

