Legal AI
JusticIA
Finds new rulings from Argentina’s Supreme Court, writes an executive summary of each and publishes it to a public Telegram channel.
- Python
- OpenAI
- Telegram
- Docker
Built in house
Tested in open waters.
These are our own products. Built in-house, launched by us, and carrying real-world load every day.
- Operations platform
In productionRandall
The bot that keeps your finances up to date.
- React
- Node.js
- PyTorch
- Restaurant management
In productionRandall Resto
Online menu, QR-code ordering, tables and kitchen flow.
- Node.js
- React
- MercadoPago
- Railway
- Billing & finance
In productionCuentasPro
Invoicing, current accounts, thermal printing and CSV import/export.
- Electron
- Node.js
- Next.js
- Railway
- Dermatology AI
In productionDermAI
Skin screening from a photo.
- Ionic
- Python
- PyTorch
- Legal AI Open source
JusticIA
Supreme Court rulings, summarized by AI and posted to Telegram.
- Python
- OpenAI
- Telegram
- Docker
Custom software
We join the crew.
We partner with founders who care about high-end execution. Same hands, same standards, from discovery to deployment.
- 01
Plot the heading
We map the problem and the destination before writing a single line of code. If the vector is unclear, we scope a paid discovery, not a guess.
- 02
Ship, then trim
A working version in the wild early, corrected on every user gust. We don't believe in big-bang launches; we believe in momentum.
- 03
Stay on watch
Launch day isn't the destination; it's just the start of the voyage. We stay on board to iterate, scale, and manage the load.
Applied AI
Instruments that read the water.
We bring machine learning and language models into the processes you already run. Measured against real cases, shipped to production, kept on watch.
- 01
Documents into data
Invoices, contracts and forms read, classified and loaded where they belong.
- 02
Assistants on your channels
Bots on WhatsApp, Telegram or the web that answer, register and escalate.
- 03
Models that predict
Classification, forecasting and image analysis trained on your own data.
- 04
Workflows with a hand on the helm
Agents that run multi-step tasks, with human review where it matters.
- 01
Take soundings
We audit the data and the process, and say plainly where AI pays and where it does not.
- 02
Trim to the wind
We integrate the model and evaluate it against real cases before anyone relies on it.
- 03
Hold the line
In production we monitor quality and drift, and correct the course as the data shifts.
Already at sea
Instruments
The Navigational Gear.
The stack we hold a course with. Predictable infrastructure where stability matters; cutting-edge speed where it counts.
Frontend
- SvelteKit
- React
- TypeScript
- Vite
Backend
- Node.js
- Python
- PostgreSQL
Infrastructure
- Vercel
- Railway
- Cloudflare
- Meta Platform
AI & data
- Claude Suite
- PyTorch
- Pandas
How we work
Trim constantly. Hold the line.
Sailing is constant correction toward a fixed destination. So is good software. We adjust without drifting.
Set a heading
Ready to catch the wind?
Tell us your destination. We'll show you the fastest line to get there.
Prefer voice? Hail us directly: Book a call
30 minutes. No pitch, just a read on the work.