CM / 001Applied AI Lab - Est. Costa Rica
AI systems for
companies that
need more than ideas.
Controlled Mayhem is an applied AI lab building client products, internal tools, open-source infrastructure, and agentic systems that move from concept to production.
/ 01 - Studio
AI product studio from Costa Rica
/ 02 - Status
Client systems in production
/ 03 - Output
Open-source tools in development
/ 04 - Team
Built by product architects and full-stack engineers
§02 - What we do
Four practices, one workshop.
S - 01AI Product Strategy and ArchitectureFor: companies with AI ideas, no clear path↓S - 02AI Product DevelopmentFor: companies ready to build↓S - 03AI Lab Systems and AutomationFor: companies that want to automate operations↓S - 04Open-Source and InfrastructureFor: developers and teams building on AI↓
§03 - The Lab
A growing portfolio of experiments and systems.
№ProjectDomainStage
L - 01 Kodus LegalLegal AI / RAGProduction↗
L - 02 CimentaConstruction AIProduction↗
L - 03 HecateDeveloper InfrastructureOpen Source↗
L - 04 DelegateMCP / API RoutingActive build↗
L - 05 TaskHiveAgent OperationsActive build↗
L - 06 MegaBrainAgent MemoryR&D↗
§04 — Principles
How we build, what we build, and who we build with.
01
Production Over Demos
We care about systems that survive real users, messy data, and operational pressure.
02
Human-in-the-Loop by Design
AI should accelerate work without removing human judgment where it matters.
03
Architecture Before Automation
Bad workflows do not become good workflows just because AI is added.
04
Open When Useful
Some internal tools should become open-source infrastructure when they help the wider ecosystem.
05
Small Teams, High Leverage
Compact systems that give small teams capabilities usually reserved for larger companies.
Pull · Q.01
"The goal is not to build AI for the sake of AI. The goal is to build systems that make work better."— Controlled Mayhem
§05 - Start a thread
Have an AI product idea, internal automation problem, or messy workflow?
Tell us what is broken, what is promising, or what is barely a sketch. We will figure out the system behind it.