DonApo

Turning a Transcript into a Testable Operating Model

A typed Events MX blueprint became the bridge between provider-assisted extraction and deterministic business behavior.

By June 2026, DonApo's voice input had become a transcript-to-blueprint pipeline. An OpenAI-backed provider could propose structured Events MX data, while DonApo retained final authority through schema validation, missing-field analysis, contradiction checks, and explicit owner or administrator review.

The next step was to make the blueprint executable without turning it into an unconstrained chatbot prompt. A deterministic state machine, Mexican Spanish input parser, quote engine, and availability engine were added. Together they could collect event details, calculate supported estimates from configured prices, reject unsupported or incomplete calculations, and require human confirmation for uncertain availability.

On 20 June, a persisted Web preview and WhatsApp-style administrative simulator made this behavior observable. It could create preview conversations and eligible lead, event request, quote, and handoff records, while retaining clear metadata that no payment, reservation, owner notification, or real WhatsApp action had occurred.

← Back to DonApo