AUTOMATIONAI · SOFTWARE · INTEGRATIONbuilt to fit

PROBLEM

Meta's official API closes a conversation 24 hours after the customer's last message. Past that, only paid templates — and the sales team has no idea the clock is running. Leads go cold in silence, template costs climb unnoticed, and nobody can say which agent let the conversation drop.

WHAT I BUILT

A full CRM on top of the official API: real-time radar of conversations about to blow the window, per-conversation cost (Meta spend plus template spend), a live ranking that rewards fast replies and closed deals while penalizing abandoned leads, multiple agents on one number with routing, transfers and unbroken history, and 39 permission levels across agent, supervisor and owner.

RESULT

A live product with three tiers (Basic, Pro, Enterprise), contact migration with history and tags preserved, a reply copilot on Pro, and ERP integration plus custom automation on Enterprise. It runs in the phone browser so the team can work from anywhere.

PROBLEM

Purchased lead lists arrive stale, with no filter for size or activity, and the outreach is identical for everyone — which guarantees a low reply rate and a burned number.

WHAT I BUILT

Queries the official public CNPJ registry with filters for activity code, company size, region and registration status, enriches contact data, and generates AI-personalized outreach per company, delivered over WhatsApp at a controlled cadence.

RESULT

A prospecting pipeline that goes from a generic list to a queue of genuinely eligible companies, with the first message already written in each business's context.

PROBLEM

The signal that a company is ready to buy is never in one place — it is spread across sources, and nobody has time to cross-reference all of it by hand every day.

WHAT I BUILT

Collectors for multiple sources, deduplication, and an AI layer that classifies intent and ranks leads by real likelihood of a conversation, not alphabetically.

RESULT

A daily lead queue already sorted by who is worth today's call.