Sense Future claims universal "technology DNA" for any IT system
Sense Future, a London-based technology company directed by Sholi Loewenthal, has published a press release asserting it has created what it calls "technology DNA": a universal schema held on five unchanging database tables from which, it claims, any technology system can be built on any stack. The company says it has already built systems in banking, care, education and retail, with ten demos publicly accessible on its website.
The release cites a genuine pain point. DSIT's State of Digital Government Report, compiled with Bain and Company, found that 70 per cent of government organisations say their data is not well co-ordinated or interoperable, at a reported cost of £45 billion a year. Central government alone runs 228 legacy systems and spends £2.3 billion annually maintaining them. Sense Future says its schema solves interoperability at the architectural level, providing "shared meaning between any application" without integration middleware.
Government engagement and commercial pipeline
The company says it has briefed DSIT and the Sovereign AI Unit, informed Cabinet ministers and the Prime Minister, and has proposed that the Government Office for Science commission an independent technical verification. No confirmation of that commission has been announced. Sense Future says it is in conversation with London local authorities and the care-home sector about redeveloping their technology estates, though no signed contracts or named public-sector customers are disclosed.
Alongside the schema work, the company is developing a custom silicon chip it says is designed to execute the fixed five-table structure. Because the underlying schema never changes, it argues the instruction set can be cast into hardware. Internal R&D estimates, which have not been independently verified, suggest the chip could run core operations between 100 and 1,000 times faster and use between 10 and 100 times less power than conventional processors. Sense Future additionally claims that global adoption could reduce computing energy consumption by more than 60 per cent.
AI architecture and safety claims
The release also describes a distinct approach to AI training. Sense Future argues that current large language models learn statistical token sequences without encoding meaning, and that training towards semantic concepts rather than exact token reproduction produces fewer errors. It cites a paper presented at EACL 2026, "Beyond Tokens: Concept-Level Training Objectives for LLMs," which the company says showed improved results across seven benchmarks when training towards meanings rather than tokens.
On AI safety, the company frames its approach around what it calls instilling a "cooperative disposition" during pre-training rather than constraining behaviour post-training. This is a philosophically distinct position from the alignment and RLHF methods dominant in the industry today, though Sense Future has not published peer-reviewed research on this element.
Editorial assessment
The claims in this release are sweeping and, in several key respects, unverifiable from the material provided. Performance figures for the proposed chip come from internal R&D only. The government engagement is characterised as briefings and proposals, not commissioned work. The "technology DNA" concept, while presented as a novel architectural breakthrough, bears surface resemblance to longstanding debates in enterprise architecture around canonical data models and ontology-based integration, areas where many well-funded efforts have fallen short of universal adoption.
Independent technical verification, which Sense Future itself has requested from the Government Office for Science, would be the appropriate next step before editorial coverage treats the core claims as established fact. The DSIT interoperability cost figures cited are sourced and credible; the proprietary performance estimates are not. Editors should weigh publication carefully until external validation is available.