A developer team at Vercel now has an agent writing a quarter to a third of its pull requests. The exact share is not the interesting number. What is interesting is that someone measured it at all.
In June an open model cracked 50 points on the leading independent AI benchmark for the first time; six weeks later a second one came within three points of the frontier. Why the average number is still almost irrelevant for your own decision.
Same prompt, same model, twice, two different answers. Not a bug: the model draws from a probability distribution at every word. What temperature, top-p, and top-k actually set.
An AI vendor can disappear no matter how big it was. Three questions decide whether that means a migration for your company or a rebuild from zero, and they belong before the contract, not after.
300 paid Copilot seats are not 300 users. What Microsoft's own reports count as "active," what the widely cited 20-30 percent figure actually measures, and why it likely understates the gap rather than overstates it.
Data "stored in the EU" sounds like protection from US access. It is not, automatically. What the CLOUD Act and FISA 702 mean, and when it matters for your company.
