Our flagship technology
At Lekton.AI we have chosen Claude by Anthropic as the reference technology for our generative AI solutions. Our products — from document platforms for legal cases to 231 compliance systems — are built on the Claude models, which stand out for the quality of their reasoning over long, complex texts, their reliability in following articulated instructions and their robustness in processing documents in Italian: exactly what is needed when the output must withstand the scrutiny of a legal professional or a supervisory body.
Claude in our products
We use the Claude models through the Anthropic API as the analytical engine of our pipelines: structured extraction from heterogeneous documents, multi-pass analysis with evidence validation, generation of formal deliverables in legal language, and reading of scanned documents through the models’ vision capabilities. The model family — from the most powerful for complex analyses to the fastest and most cost-effective for high-volume tasks such as transcript correction — allows us to assign each step of the pipeline the model with the best quality/cost ratio.
Claude Code: how we build software
There is a second, less visible but equally strategic layer: Claude Code, Anthropic’s agentic development tool, is an integral part of how we build software. We use it daily to accelerate the development of our products, to refactor complex codebases and to migrate legacy systems. It is also a service we bring to our clients: we help development teams adopt AI coding tools in a structured way, with measurable productivity results.
Agentic AI with Claude
The Claude models are today the reference for agent-based architectures: systems in which the AI does not merely answer, but orchestrates tools, queries document repositories, performs checks and produces complete artefacts. Our document analysis pipelines are built precisely on this paradigm — with one non-negotiable principle: every output declares how it was produced and remains subject to validation by the professional.
Transparency and reliability
We also choose Anthropic for its approach to model safety and reliability, consistent with the regulated sectors we operate in. And we practise the same transparency: part of the basic components of our platforms is published open source on our GitHub.
We practise what we propose
Transparency also applies to this website. This website and our open source code repositories are built and maintained with the support of Claude. The content, the organisation of the pages and the editorial updates go through Claude as an editorial and site-management assistant; the code published on our GitHub is born and evolves with Claude Code. It is the most direct way of showing — on our own work — what it means to integrate these tools into a real workflow: with a person always deciding, reviewing and validating the final output.
Want to find out what the Claude models can do for your document processes? Contact us.