Quesma Blog
quesma.com · Company blog · on HN since 2023
Company blog of Quesma, a database and AI tooling startup, covering AI agent benchmarks, LLM costs, and analytics infrastructure.
What HN saysThe founder comments on it directly, and HN readers question benchmark details such as the telecom agent policies. HN links it for AI benchmark and tool-use posts.
15HN people shared it
3linked it in comments
72stories submitted
820story points
2 months agolast shared
2023people per year2026
Most-shared posts
- 1Nano Banana Pro: raw intelligence with tool use3 people · 12 points · first shared 2025 · 2 comments on HN
- 2Tau² Benchmark in Action: Early Results and Key Takeaways2 people · 16 points · first shared 2025 · discussion on HN
- 3We hid backdoors in binaries – Opus 4.6 found 49% of them2 people · 8 points · first shared 2026 · 1 comments on HN
- 4AI for coding is still playing Go, not StarCraft2 people · 8 points · first shared 2025 · 2 comments on HN
- 5Don't Let Apache Iceberg Sink Your Analytics: Practical Limitations in 20252 people · 7 points · first shared 2025 · discussion on HN
- 6Tokenflation: When "Hi" triggers 33 tool calls2 people · 6 points · first shared 2026 · discussion on HN
- 7Baba Is Solved by Fable 5 and GPT-5.6 Sol, but at what cost?2 people · 5 points · first shared 2026 · 2 comments on HN
- 8The true cost of saying "Hi" to an AI agent2 people · 4 points · first shared 2026 · discussion on HN
- 9BinaryAudit: Can AI find backdoors in raw machine code?2 people · 2 points · first shared 2026 · 1 comments on HN
- 10Which chart would you swipe right2 people · 2 points · first shared 2025 · discussion on HN
- 11Tau² benchmark: How a prompt rewrite boosted GPT-5-mini by 22%1 people · 197 points · first shared 2025 · 65 comments on HN
- 12The security paradox of local LLMs1 people · 160 points · first shared 2025 · 87 comments on HN
- 13CompileBench: Can AI Compile 22-year-old Code?1 people · 148 points · first shared 2025 · 65 comments on HN
- 14Qwen 3.6 27B is the sweet spot for local development1 people · 117 points · first shared 2026 · 68 comments on HN
- 15I burned all my tokens researching how to save tokens1 people · 11 points · first shared 2026 · 4 comments on HN
What HN people wrote when they linked it
jakozaur · 2025-03-19
Kibana is great, but managing Elasticsearch is pain. Slow ingestion, query performance, takes a lot of space, harder to tune up.
Modern columnar SQL such as ClickHouse are 10+ times more efficient in real-world use cases.
I'm a CEO and founder of Quesma, which, let's use Kibana with ClickHouse: https://quesma.com/
Forever free, source-available license.
stared · 2026-08-01
"A Visualization Language for the AI Era", as a tagline, sounds weird.
I had the best success with popular and versatile packages like matplotlib and ggplot2 - even 1.5 ago (vide https://quesma.com/blog/which-chart-would-you-swipe-right/).
Now, frankly, my go-to visualization package is React, sometimes with a pinch of D3.js (e.g. https://p.migdal.pl/tree-of-tree/).
blndrt · 2025-09-17
Thank you for the feedback!
In this (telecom) benchmark you can review agent policies and manuals here: 1) https://github.com/sierra-research/tau2-bench/blob/main/data/tau2/doma… 2) https://github.com/sierra-research/tau2-bench/blob/main/data/tau2/doma…
Of course these are just parts of the prompt, you can inspect benchamark code to see how these are rendered to actual LLM calls.
In case someone is not familiar with framework methodology I've wrote a separate article covering that (with some of my thoughts) -> https://quesma.com/blog/tau2-from-llm-benchmark-to-blueprint-for-testi…