The idea for Broco was born out of my own stubbornness. Everyone was throwing AI into the cloud, and I went the other way.
Why? Because the companies that need AI the most have a legitimate fear: sending their data outside. Operational data is strategic. Would you send yours?
Here I'll tell you the path I chose and what I learned from it.
The harder path
I decided to make AI run inside the company, on-prem and at the edge, with a local model. It automates processes and talks to ERP, CRM and sensors without the data ever leaving.
It took more work. And that is where Broco came from.
What exists today
BrocoEngine is the agentic engine. It has 27 native skills and orchestrates models, tools and memory with a local LLM.
BrocoBot is the customer service and sales chatbot. It is the one answering on our website.
Broco PC I is the Brazilian hardware that runs all of this inside the company. And brocoAds builds the entire marketing structure and delivers results, without working as an agency, already with real clients.
Am I being paranoid?
I don't think so. Research shows that a share of what employees paste into public AI is confidential data.
And regulation is tightening. The EU AI Act already brings obligations for high-risk systems starting in 2026. Deloitte also records companies racing toward on-prem and agentic AI.
Those who take data seriously are bringing AI in-house.
What I learned
Hard technology is not the bottleneck. The bottleneck is trust.
You don't sell the model. You sell the peace of mind that the data did not leak and that the number went up.
And you, would you let your operation's data leave the building to get AI?
