LUIS ROSALPRODUCT EVIDENCE
← All workINTERACTIVE DEMOAI Operations

AI OPERATIONS · SALES & SUPPORT

AI that can answer.
Operators stay accountable.

This synthetic demo shows the operating pattern behind a WhatsApp-first product I designed and built: grounded responses, lead qualification, explicit confidence, and human handoff.

ROLE
AI product & full-stack engineering
FOCUS
Reliability, operator control, auditability
DATA
Synthetic; no client or customer records
LIVE DEMO

Select a conversation. Inspect the cited answer and confidence, then approve it or route the case to a human.

OPERATIONS INBOX

Needs review

● LIVE
MO

Mara Ortega

Norte Retail · WhatsApp

Awaiting review
CUSTOMER · 09:42

We have six locations. Can each store use its own catalog while HQ sees every conversation?

✦
AI RESPONSE · DRAFTQualified · multi-location
CONFIDENCE94%

Yes. Each location can operate with its own catalog and routing rules while headquarters retains a consolidated reporting view. I’d recommend a short setup review to map locations, roles, and escalation ownership before import.

↳ KB-14 Multi-location↳ OPS-07 Role boundaries
THE PRODUCT PROBLEM

A chatbot is easy. A controlled operating system is harder.

Useful automation needs business knowledge, tenancy boundaries, fallback behavior, evaluation, reporting, and a visible path for people to intervene.

WHAT I OWNED
  • Product workflow and admin experience
  • AI response and qualification behavior
  • Knowledge, evaluation, and audit surfaces
  • API services and staged deployment workflows
THE DECISION PRINCIPLE

Automate the repeatable response. Escalate uncertainty before it becomes a customer-facing failure.