AUTOMATING REI LEAD QUALIFICATION WITH AN AI VOICE AGENT
Designed an AI-powered phone agent that qualifies inbound leads in real time — reducing manual effort and filtering out ~80% of low-quality inquiries.
Year
2025
Field
Real Estate
Client :
Pebblerei CRM
Category
Feature

PROBLEM
The team relied on manual calls to qualify inbound leads, which was time-consuming and inconsistent.
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |
WHAT I DID
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |
More Projects
More Projects
AUTOMATING REI LEAD QUALIFICATION WITH AN AI VOICE AGENT
Designed an AI-powered phone agent that qualifies inbound leads in real time — reducing manual effort and filtering out ~80% of low-quality inquiries.
Year
2025
Field
Real Estate
Client :
Pebblerei CRM
Category
Feature

PROBLEM
The team relied on manual calls to qualify inbound leads, which was time-consuming and inconsistent.
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |
WHAT I DID
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |
More Projects
More Projects
AUTOMATING REI LEAD QUALIFICATION WITH AN AI VOICE AGENT
Designed an AI-powered phone agent that qualifies inbound leads in real time — reducing manual effort and filtering out ~80% of low-quality inquiries.
Year
2025
Field
Real Estate
Client :
Pebblerei CRM
Category
Feature

PROBLEM
The team relied on manual calls to qualify inbound leads, which was time-consuming and inconsistent.
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |
WHAT I DID
Designed an AI voice agent integrated with Retell AI that:
conducts real-time conversations with leads
captures structured data during calls
routes qualified leads to the team

KEY CONVERSATION DESIGN DECISIONS
Designing the AI agent required balancing natural conversation with the business need to collect structured qualification data. The challenge was not only defining what the AI should ask, but also how it should handle uncertainty, build trust, and guide users toward meaningful outcomes.
Challenge | Design Decision |
|---|---|
Unpredictable responses | Introduced fallback paths and clarification prompts |
Low user trust | Clear AI introduction and transparent messaging |
Missing qualification data | Progressive follow-up questions |
Ambiguous answers | Confidence scoring and routing logic |
AGENT TESTING AND QUALIFICATION FLOW
UX flow defining how users configure, test, validate, and review AI-generated lead qualification results before deploying the agent.
Step 1 - Configure the agents test prompt

Step 2- Agent Connects

Step 3 - Start AI conversation and generates responses

Step 4 - Review qualification results -> Refine prompt & repeat

PRODUCT OUTCOMES
Challenge | Product Outcome |
|---|---|
Manual qualification | Automated AI qualification |
Inconsistent notes | Structured CRM data |
Sales reviewed every lead | AI filtered ~80% of non-relevant leads |
Time-consuming handoff | Faster routing to Sales |
Difficult to scale | Reusable AI qualification framework |




