Your Experts. Their Knowledge. One System.
At Datamantics, we develop solutions that enable people from different disciplines to work efficiently together through targeted digitalization tools.
People carry enormous amounts of case-based knowledge and experiential insights in their heads—knowledge that is incredibly difficult to externalize. We help encode this knowledge into software solutions that can be easily shared and applied.
What matters is the time of your experts.
What matters is making useful what they know.
Technology serves that goal—not the other way around.
Knowledge as Software
Your knowledge is your value, not your technology.
We strive to make AI trustworthy and useful. We understand that the use of LLMs—despite their enormous value—represents a paradigm shift from deterministic to stochastic. Therefore, we apply techniques that work under this premise.
We aim to commoditize technology as much as possible, especially LLMs, to make our users and clients less dependent on individual service providers. We offer tools that enable them to fully exploit their own knowledge. We use strategies and standards that make information accessible to both humans and machines—with clear sources of truth and end-to-end traceability.
Our Philosophy
Your knowledge is your value, not your technology.
Knowledge Wins
Technology changes constantly. Your accumulated knowledge and experience stay. We help you capture what matters.
Tech is Commodity
We commoditize technology—especially LLMs—to reduce dependency on providers. Your knowledge remains yours.
Virtuous Circle
Knowledge improves as your company improves. Stay current with tech changes. Stay resilient through people changes.
Trustworthy AI Reports
Automated report generation with traceable sources.
You know the problem: AI tools generate impressive reports - but where do the numbers come from? Can you rely on them?
We deliver automated report generation with complete source traceability. Every statement is verifiable. No guessing. No hallucinating.
The naive approach fails because errors accumulate mathematically across data points, there’s no intermediate validation, and cascading failures pollute entire outputs.
Our solution separates extraction (AI answers atomic questions) from assembly (deterministic rules combine facts). The result: full auditability, reproducibility, and hallucination mitigation through consensus voting.
SmartPreop
AI-Powered Preoperative Assessment Support — in collaboration with Consorci Hospitalari de Vic (CHV)
Anesthesiologists review extensive patient documentation before every surgery — medical history, lab results, imaging, medications. This takes 30–40 minutes per patient.
SmartPreop extracts relevant clinical information from patient records (in Catalan and Spanish), organizes it following ESA 2018 guidelines, and presents it for physician review. Every extracted data point links back to its source document.
The physician stays in control. SmartPreop extracts documented facts — it does not generate clinical conclusions. All final assessments require physician validation.
What Our Clients Say
We struggled with BOQ imports — every general contractor sends different formats, and we needed them all in one system. We tried letting the LLM transform the documents directly, but it was expensive and error-prone.
Albert showed us a smarter approach: let the LLM write the code that does the transformation. Suddenly we had something testable, reliable, and cost-effective.
That shift in thinking solved this long existing problem and we can apply to a lot of new ones. Highly recommended for anyone trying to use AI practically, not just experimentally.