We have built observability systems, optimized cloud infrastructure, maintained OpenTelemetry, and designed reinforcement learning environments for frontier AI labs. That work took us deep into agent trajectories: separating real progress from retries, broken tools, and expensive activity that goes nowhere.
Systems engineers learning how
intelligence behaves.

Quesma is building a trusted record of what coding agents do, what they cost, and where they can work better.
We created CompileBench, OTelBench, and BinaryAudit, contributed trajectory tooling to Harbor, and independently audited Blitzy's SWE-Bench Pro result. Explore our research
Quesma is venture-backed ($2.5M pre-seed) by Heartcore Capital, Inovo, Firestreak, and angels including Christian Beedgen, co-founder of Sumo Logic.
Meet our team

Jacek Migdał
Spent 10 years at Sumo Logic: joined early (~20 employees) in SF, opened the Poland office in 2015. Helped scale to $300M ARR and a $1.7B+ valuation. Also a NVIDIA CUDA and Facebook alum.

Przemek Delewski
Observability veteran with 15 years at Dynatrace and Sumo Logic: engineer, tech lead, team lead, architect. OpenTelemetry maintainer, ex-PHP maintainer. Once built navigation software at TomTom. Designs programming languages for fun.

Piotr Grabowski
Led the driver and connector ecosystem at ScyllaDB, powering infrastructure at some of the world's largest companies. Co-authored Cassandra vs. Scylla benchmarks (spoiler: Scylla won). Enjoys low-level programming, performance optimization, and wrangling big datasets.

Cezary Piwowarczyk
Seasoned generalist with 10 years' experience in driving diverse teams and projects. Helped scale Codility from startup to mature SaaS. Thrives at the intersection of tech and business.

Przemysław Hejman
Over 15 years of industry experience. Spent almost six years at Elastic building its SaaS offering, spanning 50+ geographic regions of all major cloud providers. Knows the joys and horrors of building large scale-infra at global scale.

Rafał Strzaliński
SQL expert with 20+ years in full-stack development. Spent a decade designing and modeling complex databases. Believes that a “dirty job” is sometimes the best job.

Piotr Migdał
PhD in Quantum Physics. Former co-founder & CTO of Quantum Flytrap. Deep learning consultant and data visualization specialist. Creates ggplot2 charts and Grafana dashboards, combining academic research with practical software engineering.

Małgorzata Piotrowska
Built a powerful expert network at GLG in London, helping investment funds and large corporations gain a competitive information edge. After 8 years in the UK, returned to Warsaw to tackle a new challenge: finding and developing the talent that teaches AI new expert skills.

Bartosz Kotrys
Full-stack platform engineer: nine years at Finder, rising to Principal Engineer and Staff SRE. Owns the whole stack, cuts cloud costs, and builds the developer experience that makes teams faster. First move: throw out the playbook.
Grzegorz Kapusta
Engineering leader, startup investor, and advisor. CEO of Polidea (acquired by Snowflake), then Snowflake site lead, growing the Warsaw office to 400+ engineers. Established Invisible Technologies' engineering hub in Poland. Now making sure AI agents spend their tokens wisely.
Our institutional investors
Help us define how AI work gets measured.
Bring us the agent work you need to understand. We’ll turn it into evidence you can trust.
Talk to the founder