less noise. more signal.
nuclai came out of a blunt observation shared by two people from inside the corporate world: finance departments and communications teams struggle with the same lack of clarity.
AI entered companies through the people who work there, not through strategic projects. Everyone found their own tool, and the noise came with it: generic licenses, gaps in confidentiality. What’s missing today is neither appetite nor technology: it’s measurement.
Our method is exacting. We read a process before implementing a single feature. Which flow actually costs time? What data moves through it? Sometimes the reading concludes that there is nothing worth automating, and we write that, with the evidence behind it.
When there is friction to remove, we replace costly software and repetitive work with AI systems built for your case. The aim is single: turn effort into capacity, and hand your teams full control of the systems that produce it.
SME scale, same standard: AI optimization isn’t reserved for multinationals. We calibrate our diagnostics for smaller, faster-moving organizations, with the same rigor.
What we build is made to be taken over: documented, readable, yours.
the pair. two disciplines, one standard
nuclai / finance
Pedro Mestre is a trained accountant who has worked in finance since 2013. His career runs from Swiss SMEs to the analytics leadership of a Neuchâtel technology company. Along the way, he has handled accounts payable, asset management and procurement, before specialising in sales commission management.
He knows corporate finance from the inside, and one pattern kept coming back: expensive software licences, often underused, while teams still reconcile figures by hand. His answer: tools built around the actual process, not the other way round.
He works from the engine room. He builds LLM environments and Python scripts, and connects them to ERP systems to automate financial processes, from month-end close to sales commission calculations. A close-tracking tool he developed made it possible to cancel the licence of the software it replaced.
He also builds analysis and forecasting models. His rule: make data reliable at the source, so every figure can be traced back to where it came from. And he hands the tools over to finance teams, documentation included, so they stay in control of their processes and their profitability.
A clean read. On the numbers.
nuclai / comms
Sofia Mestre began as a technology journalist, then spent nine years as a project manager at a Swiss language service provider, working for luxury watchmaking and machine-tool brands.
She has seen press relations from both sides. As a journalist, she received releases that had nothing to offer her readers. As a project manager, she understood why: they are written from the brand’s point of view rather than from the angle a given publication would cover. Her answer is a prototype built on natural language processing (NLP) that adapts one release to the editorial angle of each target journalist.
She knows language technology from the engine room. For more than four years, she ran neural machine translation in production. She worked with the in-house developers to improve the tools, and edited the system files needed to deploy adapted solutions internally. She learned where these engines are dependable, and where they need a firm hand. In international communication, she keeps machine translation on-brand by enforcing terminology glossaries and tone-of-voice guidelines, so a brand’s identity holds across languages and volumes.
She also builds marketing automation: content calibrated for each social platform, and dynamic newsletters.
And she works on generative engine optimization (GEO), so that AI engines find a company’s voice, understand it and quote it accurately.
Sofia backs this field experience with formal training: a diploma as AI Specialist for Marketing and Communication, and an AI Practitioner certificate, both from SAWI, the Swiss training centre for marketing and communication.
A clean read. On the message.
How we work
We work from the canton of Bern, Switzerland, with the discretion the work requires.
No judgment, results
We don’t impose a moral view of AI. Whether your goal is a sharp cut in costs, hyper-scalability or an internal code of practice, we execute your strategy, in strict confidence.
Yours to run
Everything we build is handed over with the documentation to run it without us: the prompt library, the skills, a user guide and the logic behind each step. The system is designed to work with the language model of your choice, or several, and to move to another one when you decide.
Effort opens doors
nuclai exists because Switzerland showed its founders that steady work opens doors. Sofia Mestre and Pedro Mestre built their careers here on that principle. They now bring it to the companies they work with: serious effort, applied where it creates real value.
See where you stand
