Rubén A. Dorta A.
For 25 years I've designed BPM engines, integrated systems and led migrations with operators' teams. Today I bring the same care to applied AI: the model handles language, and the code applies the rules.
What I've learned in 25 years of enterprise systems, now applied to AI.
I wanted to show that AI can make automated decisions that follow the rules and go beyond a chatbot.
A bank dispute assistant in Spanish and Portuguese. The AI understands the request; the code applies the policy and asks the customer to confirm.
AI for language · rules in code · tested before release
Many companies keep their knowledge in documents with no semantic search. I wanted a RAG anyone can actually deploy, starting with my own notes.
Answers questions about your Markdown notes and cites the file and lines behind each answer. It runs on your computer.
Postgres · pgvector · Ollama · MCP
Order flows cross many systems and get complex fast. I want to see them clearly without touching what already works.
Follows each order from purchase to working service and shows where orders get stuck and what each failure costs. It is read-only and runs on the auditor's computer.
In development · milestone M0 closed
Telecom OSS/BSS since 2001, the last ten years at Ericsson with Claro Colombia and Entel Chile. Since 2026 I've focused on applied AI in Python, with the same principle: clear rules and results you can verify.