Closivo
From lead response to a complete qualification workflow
I led discovery and translated the findings into product scope, a bilingual PRD, a working prototype, integration decisions, and an AI evaluation framework.
The problem and my responsibility
The venture is developing a voice-AI product for UAE real-estate agencies. The initial idea emphasized responding to leads quickly, but the product needed a clearer account of what happens after that response.
A quick response is not enough when qualification is inconsistent, follow-ups are missed, and agents receive incomplete context. Automation also has to respect consent, contact limits, data boundaries, and the agency's operating model.
What I owned
- Discovery, product scope, prioritization, and founder decision workshops
- Bilingual PRD, prototype, integration contracts, and dependency sequencing
- AI evaluation, adversarial and bilingual scenarios, release criteria, and KPI design
Users, constraints and success criteria
High-volume real-estate agencies, their sales teams, and people enquiring about properties.
A qualified opportunity reaches the right person with usable context, within an agreed handoff SLA and an acceptable cost per opportunity. These are defined measures, not reported production results.
Unsettled scope
The team needed binding product decisions before committing to implementation.
Integration dependencies
Telephony approval, CRM ownership, and vendor capabilities affected sequencing and failure behavior.
Sensitive conversations
Consent, tenant isolation, retention, and escalation needed explicit release criteria.
The decisions that shaped the product
Prioritize the whole handoff
The scope followed qualification, follow-up, and human handoff rather than optimizing first-response speed alone.
Use a configurable core
Assisted onboarding and a configurable core supported agency differences without treating every customer as a separate product.
Keep providers replaceable
Vendor evaluation and replaceable adapters reduced dependency on a single provider. Integration failures, duplicates, and conflicting updates had defined behavior.
From decisions to delivery
Reframe the problem
A 42-participant industry survey changed the scope. Founder workshops settled eight binding decisions in a single session.
Make the scope inspectable
I defined 100 traceable requirements in a bilingual PRD and a nine-workspace prototype before backend development.
Sequence delivery and evaluation
Mock integrations allowed work to proceed ahead of telephony approval. An evaluation framework connected 339 source acceptance checkpoints to pass criteria.
System and integration detail
Shared context
Defined shared journey context across voice, WhatsApp, and planned email, including memory and CRM ownership.
Human escalation
Specified when the system should stop, preserve context, and hand the conversation to a person.
Release gates
Consent, contact limits, security, tenant isolation, retention, and operational readiness are release conditions.
Evidence from the work
Decisions tied to evidence
The survey reframing, eight founder decisions, and requirements traceability document why the scope changed.
339 acceptance checkpoints
The framework defines traceable pass criteria for adversarial inputs, bilingual conversations, and failure scenarios. It is a test design artifact, not a claim that a production system passed every check.
Outcome and lessons
The work moved an ambiguous idea into closed discovery, agreed product decisions, a prototype, and delivery and evaluation plans. The platform remains pre-launch.
No production adoption, revenue, or passed-test count is claimed. Email is planned. Client documentation and private implementation details are not published here.
- Discovery matters when it changes a decision, not just when it produces a document.