Why I built this
Nine years on the commercial side of large organisations, watching companies fail to find things they already knew. SemanticOS is the attempt to model that properly.
From 2015 to 2024 I worked the commercial side of large organisations: B2B sales, account management, technical account management, customer success. You sit in a meeting where a team rebuilds an answer that already exists three tools away, and then you watch it happen again the following quarter.
In a siloed company everything depends on someone remembering to tell someone else. An account manager takes the call from a furious customer about work an engineer finished three days earlier, and budget goes to the wrong place because whoever knew never found out in time. Those look like communication failures. They are retrieval failures, and a graph is the shape that answers them.
The first graph I built at scale was at a telecommunications company of about five hundred people: fifteen thousand Slack channels and conversations, recorded meetings, a base of more than ten thousand international clients, close to two million quotes a year. I was also the one arguing for it before it existed as a project, because I had spent years watching what the alternative costs.
- 15,000
- Slack channels and conversations
- 10,000+
- international clients
- ~2M
- quotes a year
- 500
- people in the company
Prior work at a previous employer. The SemanticOS figures are on the evidence page, and they come from its own reports.
Since 2025 the job has been the same problem in a different seat: modelling complex data into Neo4j knowledge graphs and optimising retrieval over them for analytical and agent use cases. SemanticOS is that work generalised, from one dataset to everything an organisation holds.
I also write, working from a running list of about a hundred and seventy papers I keep coming back to. The method is always the same: take one, often forty pages of maths and physics, and cut it to something a non-specialist will finish without removing what is surprising about the result. That produced sustained conversation with people working in data, AI and audio, and it is part of how the current role came about.
Background
Data engineer, Neo4j certified. CS50 from HarvardX, nine months of university computing coursework, and a list of cloud and ML certifications. Spanish native, English C2.
If you are building something like this
SemanticOS is an independent project, paused between phases while I look for the resources and the people to carry it further. The problem it addresses does not pause with it: an organisation’s knowledge has to be modelled before anything built on top of it can be trusted. If your team is working on that, for retrieval or for agents, I would like to hear about it.
Goes to info@semanticos.io, which I read myself. Happy to give a live walkthrough of the system and the reports behind it.