Products
Tools we're building for the market
Beyond bespoke projects, we are developing vertical products that solve recurring problems in AI and data adoption.
Coming soon
Kiln
Ready-to-use data to specialise your models
Companies are starting to specialise their language models on vertical domains: internal code, regulatory documents, product knowledge bases. The bottleneck isn't the model — it's finding quality data in a usable format.
Kiln is under development. If you are working on SLM fine-tuning and want to be among the first to access it, get in touch.
Curated datasets for fine-tuning
Clean, validated and formatted data ready to specialise Small Language Models on specific verticals. No preprocessing pipeline to build from scratch.
API access
Simple integration with any existing training workflow. Download samples, filter by domain, access datasets in standard JSONL format.
Integrated evaluation
Dashboard to measure fine-tuned model performance against the baseline. Objective metrics, standard benchmarks, comparable results.
Interactive tool
Fermata
From theory to practice: not just a case study, but a tool you can try
Fermata is the project where we applied data engineering, multi-factor scoring and a Claude-based conversational agent to analyse Bologna's public transport network.
Explore the interactive map and ask the agent — the data and answers are real.
Discover Fermata →9,563 stops analysed
The entire TPER stop network in Bologna, evaluated with a multi-factor score to spot redundancies.
Interactive map with real data
Filter stops by recommendation and score, and explore the equity layer on hexagonal cells.
Claude-based conversational agent
Ask natural-language questions about the candidates and read the agent's real answers, with cited sources.
We're building Kiln with early users
If you are evaluating SLM fine-tuning for your business, we want to talk to you. Your real needs drive the product.
Leave your email for early access