Trismik Secures £2.2m Pre-seed Funding
Science-grade LLM evaluation startup Trismik quietly raises £2.2M to transform how AI capabilities are measured, with the company’s unique approach to adaptive testing allowing AI builders to go from test to insight in seconds rather than minutes or hours
CAMBRIDGE, UK - 24th September 2025 12:00PM UK - While AI labs race to build more powerful models, a fundamental problem threatens progress: we’re no longer able to meaningfully measure what these systems can actually do. Traditional benchmarks have become saturated, with multiple models scoring above 90% accuracy on popular benchmarks like MMLU and GSM8K, creating a challenge for businesses that want to measure and adapt the ability of their models to perform a task and communicate results with other stakeholders.
Generic public benchmarks often poorly reflect the domain-specific requirements and proprietary data distributions that enterprises require to continuously evaluate models, while new regulatory frameworks demand particular assessment types for compliance. The result is a perfect storm - teams lack both the granular capability measurements needed for model selection and the ability to adapt tests over time to safeguard deployment, while racing against competitive pressures demanding ever-faster development cycles.
However, one solution provided by a team from University of Cambridge, one of the top centres for AI internationally, has opened up an interesting debate about the right approach for evaluating the next generation of models. Newly formed underneath the company Trismik, Cambridge scientists are proposing using proven psychometric methods for measuring human intelligence in combination with adaptive tests that adjust difficulty in real-time.
Trismik’s team blends former engineering and commercial leaders from Amazon and Salesforce with Cambridge scientists, its breakthrough - to apply Item Response Theory (IRT) and Computerized Adaptive Testing (CAT), the scientific foundations of standardised human intelligence tests, to AI evaluation.
Just as educational psychologists measure human intelligence by adapting question difficulty in real-time, Trismik's platform adjusts AI evaluation complexity to precisely map model capabilities. The company has confirmed the £2.2m in pre-seed financing was led by Twinpath Ventures, with participation from Cambridge Enterprise Ventures, Parkwalk Advisors, Fund F, Vento Ventures and angel investors from Ventures Together.
"If we want to trust AI, our methods have to be as rigorous as our ideas. Benchmark saturation is creating problems in every domain, from general knowledge, to reasoning, math, and coding. Scientists, researchers and technical teams face mounting pressure as evaluation is exploding in importance and has become essential for tying AI to trust. We need an evaluation framework that scales and can support this."
— Professor Nigel Collier, a Cambridge NLP researcher and Trismik's Chief Scientific Officer
Collier, who started his career in the 1990s with a PhD in machine translation using neural networks, has been increasingly focused on how we can ensure AI acts as a trusted partner to humanity rather than a risk to it - and is among the most prolific researchers globally, having published over 200 papers. Collier’s curiosity for whether AI could be assessed in the same efficient and fair way as humans, created the genesis for Trismik’s proprietary approach to adaptive evaluation.
In 2023, when Collier met co-founder Rebekka Mikkola, a repeat founder and enterprise sales executive with a passion for building in AI and opening doors for women in tech, the pair were backed early by Cambridge Enterprise, and, in a design partnership with a major UK telco, they built Trismik’s first MVP. In 2025, former Amazon scientist and TEDx speaker Marco Basaldella joined as CTO, completing a founding team that blends a formidable combination of science, engineering, and commercial expertise.
Trismik’s unique approach, if successfully translated, could fundamentally change how we think about model capabilities - moving from crude accuracy percentages to nuanced ability distributions. For organisations spending $ 100,000+ monthly on GPU compute for evaluation, Trismik's approach could reduce evaluation costs by up to 95% and provide more granular insights. Adaptive tests deliver near-identical rankings to full evaluations: on 4 of 5 datasets, Spearman correlations exceed 0.96 - while requiring just 8.5% of the questions.
Having finalised its raise, Trismik will now focus on launching its product for AI builders, providing its classical and adaptive evaluation with a growing number of popular public datasets covering factuality, alignment, safety, reasoning, and domain specific knowledge, giving scientists access to broad benchmarks for interpreting model performance.
Today, the company’s LLM Experimentation Platform delivers lightweight, accessible tools for fast and reliable evaluation. Looking ahead, Trismik’s vision is to expand the platform into a comprehensive environment for designing, running, and analysing LLM experiments, including fine-tuning and prompt engineering. Trismik will begin onboarding users towards the end of 2025 - early access available through the company’s website.
As Trismik continues to develop advanced experiment tracking capabilities, richer data visualisation, and prepares for enterprise-level compliance, its founders anticipate the launch of its enterprise solution early in 2026.
"The AI evaluation market is at an inflection point. Every AI team we speak with is drowning in evaluation overhead, it has become the hidden bottleneck preventing teams from shipping faster and with confidence. Trismik's approach is compelling because it applies proven scientific methods from a completely different domain to solve this problem. When you can reduce evaluation time by two orders of magnitude while actually increasing measurement precision, you fundamentally change what's possible in AI development cycles.”
— John Spindler, Twinpath Ventures
"Trismik exemplifies Cambridge’s continued contribution to global AI development with the team combining world-class academic credentials and practical industry experience that has given them the unique authority to define how AI capabilities should be measured. By solving a pivotal challenge in AI adoption, Trismik is positioned to drive trust at scale - we’re excited to support their journey to market."
— Dr Christine Martin, Head of Ventures at Cambridge Enterprise