Contracted Intelligence Coop
The industry traded quality for speed. We're buying it back
Target: To establish a software cooperative, that effectively uses Generative AI to increase software quality not quantity.
- To use AI as a real competitive advantage, not a fashion statement
- To sell mostly to international markets
- To seduce the engineers that left a FAANG because they hated the politics. The kind who wants skin in the game and a voice in decisions.
Why a cooperative?
Employees are seeing the adoption of AI technologies as threat, They know that there is a wave of firings because of AI gains in productivity. And no-one likes to train or develop their replacement. The people that are happy with the economic efficiencies of AI are the owners of the company. But in a cooperative the employees are also the owners. So it is in a cooperative's best interest for members to increase productivity with AI. Because they are also participants in the economic benefits of AI.
That is also why an AI native cooperative is better. Because other coops and other companies are currently thinking about downsizing. But a coop that is born with AI efficiencies already taken into account is going to be small from origin. No need to share profits with more people than necessary.
Since October, 2022 when ChatGPT and later Claude came into the market, software companies have been claiming that age of programming is over. For example Anthropic the company behind Claude say that 90% of their code is written by Generative AI. Microsoft claims that 30% of their code is written by Generative AI.
But this has had the negative effects that:
Amazon called an emergency meeting after outages caused by AI-assisted code changes. The fix: put senior engineers back in the loop – This week in AI news
On one end, the Anthropic team is a massive user of AI to write code (80%+ of all code deployed is written by Claude Code). They ship amazingly fast. On the other hand, seeing these beyond terrible reliability numbers suggests there might be a downside to all this speed. – gergelyorosz
Best summarized by:
All of this AI speed is the opposite of what is needed. Bring me 10x quality or get the f*** out. – zate75
On this cooperative we are going to use Generative AI to create 10 times more quality not more code.
To achieve that we will use the branch of programming called formal methods, in particular the "Contracted Intelligence" approach developed by Alejandro (one of the founders) for his ongoing PhD Thesis on Formal Methods and Generative AI.
Financial
To launch this project we anticipate two phases:
- Finish the Research Phase (6 months)
- This Stage started in 2025. And it is almost finished. By the end we will have several libraries created with "Contracted Intelligence" approach. So libraries for web-services, distributed systems verified by formal methods and programmed by AI.
- Develop the 1st commercial application (6 months)
- With the strong approach created on the Research Phase. And the libraries for concurrent systems. Deliver the 1st commercial product, that will exhibit better quality than ever before. We need your help to find us this 1st commercial customer
Each one with a cost of MXN $4 Million We already have MXN $2 Million and we are looking for the remaining MXN $2 Million.
Status
Since May 2025, the Contracted Intelligence approach has produced working prototypes for formally-verified web service contracts and a concurrency model for distributed systems. Both will be presented at VERIFAI-2027 in January-2027. The research phase is complete enough that the remaining six months are about hardening these prototypes into production libraries, not starting from scratch.
The founding team
Dr. Carlos Lara Alvarez
Director, CIMAT Zacatecas (Mexico's national mathematics research center). SNI Level I. 537 academic citations. Industry collaborations with Intel and commercial software firms on safety-critical and assistive technology systems. Research focus: probabilistic models, software testing, human-machine systems.
Agustin Rumayor, Ms. in Software Engineering
Staff Engineer who has been developing products with GenAI since the beginning. All the problems the industry is seeing now, he already fought a year ago. He has two GenAI products in production: rewards.codes (a consumer rewards platform) and a calendaring assistant built on AI agents.
Alejandro Garcia, Ms. in Software Engineering
PhD student, created the "Contracted Intelligence" approach that is the application of formal methods to Generative AI. In fact this coop will be the commercial application of his research.
What we are looking for
- Capital
- We model our investor relationship on Stocksy United, a photographer cooperative with 1,000+ members where investors receive 5% of company profits. We are offering the same: $2M MXN for 5% of annual profits, senior to member distributions.
- Deal Flow
We are looking for industries where correctness is non-negotiable: aerospace, medical devices, financial clearing, defense. Using your contacts find us projects where the complexity comes from the domain not the timeline. Those are our first clients.
Deal Flow partnerships are open to non-investors as well.
Alejandro García F. agarciafdz@gmail.com