Contact
How to reach us
EvalGlass is built in the open by Syntelesis-Lab, a multi-agent AI-safety lab. There is no contact form and no ticket queue — just the right channel for each thing, and a real inbox for everything else.
Who leads it
EvalGlass is led by Laurentiu Meirosu, AI Safety Researcher at Syntelesis Lab. His work spans regulated, high-stakes and mission-critical environments where the behavior of AI systems had to be evaluated, explained and understood — not simply deployed, and EvalGlass is built in that discipline.
His AI-safety research is on the safety of multi-agent AI systems. It translates directly into EvalGlass — extending its failure discovery toward the emergent failure modes and collective behaviors that arise when AI agents interact at scale, the latent, path-dependent failures no single-agent implementation can foresee.
Pick the channel
- A Core bug or feature. Open an issue on GitHub — it is the fastest path and it stays public. Open an issue →
- A change to contribute. Clone the repo, run the checks, and send a pull request. How to contribute →
- An Intelligence Investigation enquiry. Email us with your application type and the failure whose root cause you want diagnosed, to scope an Intelligence Investigation. Intelligence is in private preview. Email about Intelligence →
- A Discovery Proof enquiry. Email us with one application and the coverage gap you suspect, to scope a bounded Discovery Proof on it. Discovery is in private preview. Email about Discovery →
- A security issue. Report it responsibly, through a GitHub advisory or the security inbox. Report a vulnerability →
- Anything else. Email the makers directly at contact@evalglass.com.
Related