Keir

[Keir @probabletrain]

porsche hasn't said what engine is in the mission s, so i simulated what it might sound like: flat-8, 8000rpm, idle matched to this clip. no samples, every sound is a cylinder firing into a pipe

insanely fun to play with :D

[冬夜雪域冰莲炎 @SnowLotusFlame]

What a sexy beastly roar

[Keir @probabletrain]

this becomes feasible when:
- your agent can run the full stack locally and use the app itself to prove a change works
- it has a good framework for reasoning about what it does and doesn't need your opinion for. both for product and engineering decisions
- your agent can watch logs, db, replays after deploying to monitor the change and create followups/reverts if needed
- your agent runs a good code review
- you personally trust the model
- your agents talk to a coordinator that manages your attention and prioritises decisions for you

we have that with rook and opus 5.5

[Eloff @eloffd]

I don't know how people say they run 20 agents at once and merge 60 PRs in a day.

That's 4 minutes per PR, assuming 4 hours of actual heads down time. Not only is that not enough time to even glance at the output, it's not enough time to even nail down the requirements first.

That's just pure BS, one way or another.

[Keir @probabletrain]

@zeeg we have good results with using jev for relevancy

[Keir @probabletrain]

@neal_k_patel looks promising, but not a replacement for our jev workflows at the moment

[Keir @probabletrain]

@MahyadGhassemi atc app has it

[Keir @probabletrain]

luna speed-pictionary
i cant draw giraffes

[Tibo @thsottiaux]

Day 2.4/

Decisions API is live. We will use this ourselves to improve the experience for everyone in many fun ways. Realtime general classifier does open some great possibilities.

[Keir @probabletrain]

@steveruizok coming home to a surprise sous-vide machine my agent got me

[Keir @probabletrain]

@dinasaur_404 @Cloudflare congrats!

[Keir @probabletrain]

here's a nice ui touch - the hovered sidebar item never moves while your mouse is over it. even as other items in the sidebar shift and move.

so you never accidentally archive the wrong item because something shifted as you clicked.

[Keir @probabletrain]

@s16h_ this is a fantastic idea

[Keir @probabletrain]

@s16h_ @MetaviewAI @typesafeai haven't been this excited since opus 4.5

[Shahriar Tajbakhsh @s16h_]

at @MetaviewAI, over the weekend we shipped @typesafeai's jev into every agent on metaview.

candidate searches in our sourcing product that took minutes now take seconds. same accuracy, ~10x faster, meaningfully cheaper per search.

but the speedup isn't the interesting part. jev let us treat intelligence like software: break each agent into its smallest semantic units, query each one, set our own thresholds, and fix bugs by adding a question — not by begging in a system prompt.

that's a different way to build, and it's now our baseline. credit to the typesafe team.

optimizing for intelligence per dollar has unintuitive consequences. intuitive to @CompleteSkeptic, of course.

[Keir @probabletrain]

@cramforce @vercel_dev @typesafeai what rate limits can you give us so we can put it everywhere in prod 👀

[Keir @probabletrain]

@zeeg @0xsakanade compaction also invalidates the cache innit

[Keir @probabletrain]

jev is insane: live intent tracking as you type
we've never before had access to judgement at this latency+cost level

[Keir @probabletrain]

@ritakozlov @Cloudflare what sort of rate limits do we get?

[Keir @probabletrain]

@steveruizok son(s)

[Keir @probabletrain]

@dacc_maxi @alexandr_wang maybe he’s talking about an actual insect

[Keir @probabletrain]

@mluggy @turbahn @Muse yeah but this facilitates you doing one step of an n step flow, rather than having to do the whole thing yourself