Loading activity…
@jaredpalmer
vp of engineering @cognition. previously: vp product @microsoft. vp of ai @vercel. creator of @v0 and @aisdk. founder of @turborepo (acquired by vercel).
AI
Healthy or Satisfactory engineers can get automatic limit increases within enterprise admins’ ceilings, Cognition says.
AI
Cognition says Devin Oncall works out of the box with Datadog, Dynatrace and Sentry.
AI
OpenAI says the feature uses your conversation and how you talk to Codex to suggest what to say next.
AI
A user pairs '$200' with the name and links to a photo, without explaining what the figure covers.
AI
A post argues that agents' gains taper off when they saturate the feedback loop available to them.
AI
A user claims many already have access and speculates that Google could release the model on October 7.
AI
A user says the prompt sets a budget and time limit, then asks sub-Devins to work in parallel.
AI
A post jokingly casts canceling the subscription as a new Turing test for AI agents.
AI
A user wrote “Spotted in EWR Terminal C 👀” and linked a photo, without naming what they saw.
AI
A user wrote “Kev-27B >>>>” in a quote post, without explaining the apparent preference.
AI
The developer says Kev-27B has a 64k-token document window, and Kev 1.0 supports the TypeSafe SDK.
AI
Stamos argues that AI could make attacks cheaper to run while defenders face high costs to scan and repair older software. He says Cognition aims to help companies find and fix flaws more affordably.
AI
A user reported that Kev 27B’s full supervised fine-tuning (SFT) run was underway on September 25, 2026.
AI
A post draws a joking parallel between Triton kernel launches and useEffect.
AI
A user calls the workflow “super powerful” but says some Devin automations and AutoReview may need to score pull requests and issues before acting.
AI
A user describes the message as a “slop DM” and says they immediately tried prompt injection “for fun.”
AI
A user argues that AI is challenging assumptions about how common agency is.
AI
Kev’s developer announced 0.8B, 4B and 9B checkpoints and a fine-tuning script usable with Modal, while noting that the switch slightly hurt local performance.
AI
The developer says the new Kev models work with the TypeSafe System One SDK after a single base_url change, and that Kev-4B can run on a 32 GB Mac.
