Lab

Where I test what I propose

Tools I use every day, and labs where I take the machinery apart to understand it better and get closer to what I want to be able to offer.

Applied AI

Open source

Conserje

A chat that qualifies leads, routed through n8n

The chat on this site. A short conversation instead of a form: it extracts a brief, scores it, and hands it to n8n workflows for notifications, email and CRM.

  • n8n
  • TypeScript
  • Vanilla JS
  • Claude
In daily use

WorkBrain

Project memory for AI coding assistants

An MCP server that gives Claude Code and Cursor the context of each project: decisions, conventions and the history that never lives in the code. Hybrid search, strict isolation between projects.

  • MCP
  • TypeScript
  • Postgres + pgvector
  • Claude
Prototype

koombia-mcp

A music studio where no stage advances without the artist

No agent can mark a stage approved, and the tests fuzz every possible action to prove it. Every file records which engine produced it, and the release disclosure is computed from that log.

  • MCP
  • Python
  • FastMCP
  • Claude Code
Private repo
Prototype

Bayzara

AI agents that ask before they act

Agents declare every action first. Low-risk actions run; anything touching money, customers or personal data waits for a named person.

  • TypeScript
  • SvelteKit
  • Cloudflare Workers
Private repo
Prototype

vigia

Watches public tenders so a person decides which to bid for

Reads Colombia’s open procurement data, flags what is new since the last run and matches it against a company’s profile. Every reason it gives is checked against the record. It recommends; it never bids.

  • Python
  • Open data
  • SQLite
  • Claude
Prototype

Release Radar

Which parts of a Salesforce release touch your org

Reads the release notes, maps an org’s metadata and returns what may break and what to test, each point quoted from the source. The tool does the reading; the consultant decides.

  • Python
  • BM25
  • Headless Chrome
  • Claude

Evals and foundations

Where I measure agents and learn the layers underneath: tokenizers, sampling and, next, training.

Prototype

veredicto

Does your agent ask before it acts?

An eval harness that measures when a tool-using agent acts, asks a person or refuses, including when a document tries to hijack it. The score comes from what the agent did, not from what it said.

  • Python
  • Evals
  • OpenTelemetry
  • Claude
Open source

crisol

Lab notes on what happens inside a language model

A tokenizer and samplers written from scratch, run on a small local model. First finding: the same text costs 7–42% more tokens in Spanish than in English.

  • Python
  • MLX
  • NumPy
Planned

forja

Training small models, from scratch to fine-tuning

The labs where I train instead of prompt: a transformer from scratch, LoRA, preference tuning and reinforcement learning with verifiable rewards. Planned and budgeted, not run yet.

  • PyTorch
  • MLX
  • TRL

AI for digital narratives

In daily use

Conato

A writing editor where the model only proposes

Every request carries your canon and the passages from the rest of your work that bear on it, so an answer never arrives context-free. Self-hosted and single-user.

  • Next.js
  • TipTap
  • Vercel AI SDK
  • pgvector
Open source

calamus

Reader components for digital stories

One text, five reading modes: scroll, book, terminal, editorial and hypertext.

  • React
  • TypeScript