Trismik

- 26/09/2025
- Pre-Seed
- $2,936,967
Trismik is a science-grade experimentation platform for evaluating LLMs. Our adaptive testing enables AI builders to evaluate models in seconds, not days. We combine scientific rigour with unmatched speed, supported by clear visualisations and intuitive dashboards.
Trismik is a Cambridge University spinout shaping the future of LLM evaluation. Our team brings experience from Cambridge, Salesforce, and Amazon - with 38+ years of research, 200+ papers, and 12,000+ citations.
- Industry Software Development
- Website https://www.trismik.com/
- LinkedIn https://www.linkedin.com/company/trismik/
Related People
Rebekka MikkolaCo Founder

I’m a second-time founder, currently building Trismik - an LLM experimentation platform for AI scientists, researchers and engineers.
Most evaluation tools rely on static benchmarks that miss how models behave under pressure. We’re building for experimentation - enabling fast, adversarial testing to uncover hidden failures and track true performance, not just surface-level accuracy.
With Trismik, teams can run evaluations up to 180x faster than traditional methods - helping them move quicker, test deeper, and ship with more confidence.
If you’re working on LLMs and want to push the boundaries of how we test them - let’s connect.