01 /
AI products & agents
Useful AI starts with product judgment. We shape the interaction, model behavior, evaluation loop, and safeguards as one coherent product.
- Agent systems
- LLM applications
- Evals
- Human-in-the-loop
Independent software studio · Brazil
From uncertain idea to working system.
MOSMN designs and builds applied AI products, intelligent agents, and the infrastructure that makes them dependable in the real world.
System map01 / 04
Strategy and implementation, in one loop.
Capabilities
Deep enough for the hard parts. Lean enough to move.
02 /
Backend & platforms
APIs, data pipelines, and cloud infrastructure designed for the less glamorous realities: latency, cost, security, and change.
- Python
- TypeScript
- Cloudflare
- Data systems
03 /
Technical product strategy
Turn an ambiguous opportunity into a testable plan: what to build, where the risk lives, and what evidence should earn the next investment.
- Discovery
- Architecture
- Prototyping
- Technical diligence
Best fit
Ambitious teams with a consequential technical question: a product to prove, a system to untangle, or an AI capability that needs to work beyond the demo.
Working model
Principal-led from first question to final commit.
You work directly with Leonardo—the person shaping the approach and writing the code. Fewer layers make room for faster decisions, honest tradeoffs, and software that reflects what your business actually needs.
- 01
Frame
Find the real constraint, define success, and cut the problem to a useful first move.
- 02
Build
Work in visible increments, test assumptions early, and keep decision-makers close to the software.
- 03
Transfer
Ship with the architecture, operating context, and documentation your team needs to own what comes next.
Start a conversation
Have a difficult thing worth building?
leomconti@gmail.comShare the problem, the context, and where you feel stuck. A rough note is enough to begin.