Hypercritical raises £2m pre-seed to generate verifiable control code using formal logic
This article covers Hypercritical, a London-based AI startup, which has raised £2m in a pre-seed round to accelerate development of models that generate control software for safety- and mission-critical systems. The funds will be used to double its engineering team and pay for cloud compute as it develops a logic-driven approach to produce verifiable, testable control code for automotive, aerospace, defence and robotics.
Hypercritical, a London-based AI startup, has raised £2 million in a pre-seed round to accelerate development of models that generate control software for safety- and mission-critical systems. The cash will fund team growth and cloud compute as the company develops a logic-driven approach that aims to produce verifiable, testable control code for industries such as automotive, aerospace, defence and robotics.
Why it matters
Generating control software for physical systems typically requires lengthy engineering cycles, formal verification and expensive testing because errors can cause catastrophic failures. Hypercritical’s approach reframes the workflow: engineers specify the tests and safety constraints a system must meet, and the company’s models produce software designed to satisfy those requirements. If the method scales, it could shorten development time and change how teams certify control systems — with implications for manufacturers and regulators working on safety standards.
Inside the product
Hypercritical describes its core architecture as logic-driven rather than probabilistic. Its system uses domain-specialised agents that independently design, verify and optimise control algorithms within mathematically defined safety limits, instead of acting as generic copilots that suggest code. Products mentioned in the announcement include:
- Hyperpilot: the flagship product said to automate the development of control software for engineering teams.
- Copilots: domain-specific agents (for example, QA engineer and systems engineer copilots) that produce immediately usable, specialised output.
- Autopilot: an unsupervised generator intended to produce software that passes all specified tests.
The company claims the architecture eliminates hallucinations and produces software that can be mathematically guaranteed to meet defined tests. That is a high bar in practice: formal verification and certification for control systems remain complex and often require substantial human oversight and domain expertise. The release notes that parts of the technology stack have already been deployed with customers, indicating early real-world validation, though no customer names were provided.
Who are the investors?
The £2 million pre-seed round was led by Join Capital, with participation from Octopus Ventures, Tiny Supercomputer Investment Company (tiny.vc) and Plug and Play. Hypercritical said the funds will primarily be used to double its engineering team and pay for cloud compute to train its proprietary models. The investor line-up combines early-stage and larger venture players that have backed other deeptech and infrastructure ventures in Europe.
If you're researching potential backers in this space:
- Explore our list of AI venture capital firms in the UK
- Browse our directory of AI angel investors in the UK
Founder perspective
The company positions its technology as a way to mechanise parts of the control-software development pipeline, moving from human-written code to formally specified requirements and machine-generated implementations. Hypercritical also stated an ambition to influence industry certification practice, including potential incorporation of its methods into ISO standards for software certification and compliance. The announcement frames this as an attempt to modernise how regulated industries approach software assurance.
The bigger picture
This raise comes amid renewed investor interest in AI applied to industrial problems, where correctness and explainability are central. The deal reflects growing interest from AI investors in tools aimed at industrial software and verification. For UK and European engineering ecosystems, the challenge will be proving that model-generated control software can meet the rigorous auditing and certification processes required by sectors like aerospace and defence. Early deployments with customers suggest progress, but wide adoption will depend on demonstrable certification pathways and regulatory acceptance.
As the company hires and scales compute resources, its progress will be worth watching for anyone tracking the intersection of AI, formal methods and industrial software in the UK and Europe.
They've invested in Hypercritical
Get to know these AI investors
Click here for a full list of 7,526+ startup investors in the UK