good summary. i think you forgot heartbeat.md which powers some autonomy.

do you think the agent admin ui mattered at all?

other contributors while i think of them:

- good timing around opus 4.6 as the default model? (i know he used codex, but willing ot bet majority of openclaws are opuses)

- make immediate wins for nontechnical users. everyone else was busy chasing cursor/cognition or building horiztonal stuff like turbopuffer or whatever. this one was straight up "hook up a good bot to telegram"

- theres many attempts at "personal OS", "assistant", but no good ones open source? a lot of sketchier china ones, this was the first western one

just you. i use default substack font with some bolding

they are actually 4 companies in one. planning to ask him about that when he comes on the pod

There's network effects and then there's core competencies. OpenAI has not demonstrated their ability to create software that is not a primary use case for LLMs. Chat is absolutely not a primary use case for LLMs, and so far LLMs have been sold as a value-add for traditional software.

The argument that OpenAI has the critical mass to dethrone Slack can be made for just about any other product with an 800-pound gorilla market leader. Windows, Office, Photoshop/Premier, Search, GMail, Figma, etc. Thus far, we have yet to see OpenAI build anything like these at scale, and there's no reason to assume their successes in the LLM space will translate.

I agree that they should build killer apps like these, because they are at extreme risk of being commoditized by smaller, better, faster genAI systems, but I don't think anything they do currently shows that they can.

"You guys are reacting like engineers" is a very wave-y dismissal of the many practical questions raised about why exactly OpenAI should expand into a product that's tangentially related (at best) to their core competency of AI.

The chain of logic in the article is explicitly spelled out as: Sam Altman said OpenAI will grow into new products -> Altman says to tell them what these products should be -> You say: Slack sucks so.... how about Slack?

I think most people, engineers or otherwise, reading the article have an understandable reaction of mostly bafflement as to why we are even talking about this, specifically, to begin with?

thank you for listening!! team works hard on it. Jeff was an absolute bucket list GOAT to have on the show and to launch the pod with the new Deep Think is just icing on the cake. Every 3-4 months people remember that Google (and Jeff) has low key been accumulating basically every advantage under the sun that all the other AI majors are struggling to pull together themselves.... and then constrained by bigcorp politics. it seems like they are figuring things out though.

author here, entire editorial is entirely human written

directory perhaps less confusing.

“NPM registry directory” sounds terribly confusing in my opinion, reads like a grammatical mishap

then maybe "npm registry search"?

how are you so sure :)

how exactly is adding a stored procedure as an agent mind bending

I can only speak for my own mind ;) but the most advanced thing I'd seen prior in this regard was Google Sheets' =AI function, which is pretty convenient (if awkward) when you want to map values to LLM output.

What I specifically found "mind-bending" about this is that I don't have a clear concept of the limits of what an agent can do. In the limit case, it's basically like an independent employee, right?. So the concept of having a dedicated person sitting on each row of my database and transactionally performing any task I can describe is ... well, it IS a bit boggling to me.

Another way to look at it is: this is an extremely powerful construct for managing fleets of agents. I trust Postgres to execute all the stored procedures I ask it to. So with this tool I can easily spin up arbitrarily many agents. And state management is very simple, because they can directly edit their associated row!

IDK, the more I think about it the more fascinated I am. I'm sure there is some open source SAAS or something that has similar semantics and can do all this more efficiently, but now I know that this is a category of thing one could potentially build/use. Pretty nifty!

ok nice reply. i think i was where you were in 2021 around doing stuff in sprocs. i think pple generally follow a cycle of going overexcited about throwing everything in the database and then going "actually the database is a pretty bad production compute environment" and re-separating concerns back to different levels.

use sprocs lightly for simple fast stateless things. every other attempt at stuffing a lot of compute into the database that i'm aware of has basically failed to gain adoption (the personal awesomeness/happiness of the guy who created it aside)

Oh, like, I wouldn't actually use this specific implementation. I used to work at a shop with thousands of lines of Oracle triggers that you had to edit inline in the web browser with no version control and I shudder to think of returning again.

I'm more interested in the data flow. exa.ai got famous for promising search with massively parallel execution of LLMs on candidate results. In practice, it's never worked that well for me, but the model is very cool and has worked for me e.g. in open source work, searching for bugs across files.

Mapping N items to "N agents with state" feels like an absurdly powerful construct to me. Maybe this is just a well-known pattern that everyone has seen already, but given how much better agents have gotten in the past year, crossing the threshold from "toy" to "arguably superhuman" on many tasks, I think it just hits different.

ok gotcha. yeah i guess my background with temporal.io got me used to "every workflow instance can have a shit ton of long running state that gets persisted and rehydrated at will". check those out if you like N:N+state, whether or not it includes agents is an impl detail

Haha, I guess we see this in reverse: I see the specific framework as the implementation detail (I use, and enjoy, temporal!) I think it's the part about automatically launching a metric ton of agents that is (to use the term again) mind-bending to me.

why?