The Mathematics Search Engine

22+ min ago   (1264+ words) A customer writes in: "my last three orders are missing." Someone has to open a database and look. We built an AI agent that does that part. It reads the ticket, forms a hypothesis, queries the customer's own systems, and…...

1+ day, 1+ hour ago   (283+ words) Matt Prater, EVP of Innovation at OuterBox, documented the build in a field report published last week. "We use the same models every agency uses, but the difference is we apply decades of judgment and compile it into a system…...

1+ week, 3+ day ago   (459+ words) A new WalkMe survey of 2,000 working Americans who use AI on the job finds 90% feel confident with AI, while only a quarter say it works on the first try. SAN FRANCISCO, Aug. 25, 2026 (GLOBE NEWSWIRE) -- A confidence trap has taken hold…...

22+ min ago   (493+ words) OpenAI has published benchmarks for GPT-6 Astra showing large jumps over GPT-5.6 Sol, a 100% score on an exploit benchmark and two zero-days found in testing, alongside an admission that the model’s reasoning is harder to monitor when it attempts evasion....

33+ min ago   (1490+ words) I wanted to build something with an LLM using my own hands. Not wire an API into a wrapper, but take... Tagged with ai, machinelearning, aws, python....

32+ min ago   (372+ words) I got burned by a client who swore my original writeup was "AI-generated." It wasn't. But the accusation cost me a week of back-and-forth, so I did what any annoyed engineer does: I ran the detectors against each other until…...

3+ day, 1+ hour ago   (266+ words) Platform usage has grown more than 10x in the past year as data teams adopt the AI-native platform to build and govern pipelines and AI agents in one environment. “AI is creating enormous demand across the enterprise, but data and now…...

45+ min ago   (567+ words) Our dataset said the average loan was 2.3 million kroner. The number that actually mattered was 255,000. Both were correct. Only one of them was true. This is a writeup of three ways a dashboard can be arithmetically perfect and still lie,…...

1+ hour, 10+ min ago   (318+ words) The opening is a reconstructed incident pattern, not a claim about one company outage. The corrected model is narrower and slightly less flattering to the demo. Detected files are observations, similar to footprints around a building rather than a signed…...

1+ hour, 8+ min ago   (322+ words) An iteration that should have touched two files was burning ~350,000 tokens, because the executor was reading the repository to figure out how things are done here. The fix wasn't a better model - it was deciding what the executor is not…...