A semantic core is a large claim. This is who is making it.
Semcore is one person. What stands behind the Platform is two decades of building data platforms in gaming, payments, financial services and the public sector, and the same problem turning up in every one of them.
Why this, and why him
Semcore is founded by Ime Akpan, who has spent more than two decades building data platforms in gaming, payments, financial services and the public sector, including as Head of Data Engineering for Vanguard UK and Europe.
It comes out of noticing the same thing in every one of those estates, a decade apart and in different industries. The problem is never the models. It is that nobody can say what the words mean. What that costs, and what to do about it, is the Platform's argument to make rather than this page's.
Ime speaks at Big Data & AI World, the Data Products and Platforms Summit, CDO Exchanges and financial transformation events, on decentralised data platforms, estates built for the AI that reads them, and product-centric data teams. He mentors, and argues for broader representation in data and technology.
East Africa's mobile money platforms
Foundational work on three of them, where real-time data systems are business critical in the plainest sense of the phrase.
EY
Large-scale transformation in Tax, Technology and Transformation, where regulatory and governance demands have to be met without slowing delivery down.
Vanguard
Data engineering at the scale an asset manager runs: lakehouse data platforms, self-serve data products, and estates made ready for the AI that reads them.
How the work is done
The pattern across those estates is the same one: organisations win when they treat data platforms and semantics as products rather than as projects. The belief under that is a plain one, which is that AI only works when the context is right, and that getting the context right means investing as much in data engineering, semantics and governance as in models and prompts.
Semantic models that describe the business
Definitions written in the terms the business argues in: customers, portfolios, products and policies, rather than tables and endpoints.
Platforms that can be found and are governed
Lakehouse data platforms and data products that are discoverable, governed, and ready for the analytics and the AI that read them.
AI that answers from an agreed definition
The business context an AI system needs in order to answer from a governed definition, rather than from screenshots and spreadsheets passed around by hand.
Work done inside your team
One person working inside your team rather than alongside it, which is the point of the arrangement. What is left behind is systems and judgement your own people carry on without him.
Shape it around your own estate
Software gets shaped by whoever is in the room. A design partnership puts your estate and your problems in that room, and the page for it sets out both sides of the arrangement plainly, including what is asked of you.