sidgarimella

Max via Cursor, should've mentioned, very possible I'm seeing more Cursor than OAI. All the subagents it picked were also Sol Max... I've seen 7-10 in one turn. Regardless the top reasoning tiers being conflated with subagent count feels double edged

I've found Sol's propensity for delegating to subagents can make it... disastrously expensive, especially with each subagent having some implicit floor on further reasoning/context gathering before action.

The base model is certainly cheaper and more token efficient etc, but on large tasks cost in some way is now n^2

imo there’s a clear greenfield to have doubled down on where cursor was before in proactively keeping devs appraised of the code that they’re generating, and bridging growing gaps between abstracted chat sessions and files/directory structures I might understand less and less

This on the other hand feels like a clear reaction to cc/codex, in a way that even kind of builds an offboarding ramp?

would sure be nice if the effort spent fighting their karma was pointed at a better frontier model

Many are saying codex is more interactive but ironically I think that very interactivity/determinism works best when using codex remotely as a cloud agent and in highly async cases. Conversely I find opus great locally, where I can ram messages into it to try to lever its autonomy best (and interrupt/clean up)

where we draw the line on agent "identity" when the models being orchestrated are generally the same 3 frontier intelligences is an interesting question indeed

I would think this idea of creating a third-party to verify things likely centers more around liability/safety cover for a steroidal increase in velocity (i.e. --dangerously-skip-permissions) rather than anything particularly pragmatic or technical (but still poised to capture a ton of value)

there’s a curve where something of a conservative middle in AI marketing stunts are held to a higher level of criticism than headlines on either side

Appreciate the feedback.

Will extend the decision table template when we can

Guilty as charged

Thoughts on /permissions from the CLI?

The wording has been updated, and we’ll review our instrumentation. Appreciate your feedback.

Curious to learn if/how folks are using the PreToolUse hook in their own setup

Very cool. Seems long running AI Agents are the new monuments.

Stumbled on to a similar concept a few years ago, I think there was a paper around some kind of proto code skills for Minecraft around then... awesome to see Anthropic pursue this direction

https://www.prefix.app/blog/is-chat-the-future

+1 hyped for an mcp that I might be able to plug zed into

I myself run five IDE's in parallel with ~ 6 terminal sessions all on Claude Code. There's a time and place for each tool. Also, thanks to my Neuralink, I have Linear and Azure MCP's wired directly into my frontal cortex at all times. This makes it really easy to keep our docs up to date without lifting a finger.

probably a rare area I fully agree with HN on– the IP here seems weak and it's not hard to swap out code editors, nothing like tearing out Salesforce or other sales-driven tooling. and idk if first mover advantage actually means much in the next 10 years given how dynamic the underlying models are.

but undeniably these cos are all a great lesson in just how much cash lies in executing first/near first

the hn cynicism is crazy but unsurprising. turns out can really easily spend your whole life obsessed with a pragmatism that betrays taking any action, however directionally correct

Love this. I’ve worked on a few projects in RPA prior and I’m losing faith in selectors. I think either direct data access like this or AI based CV are the automation arms of the future.

we write about this!

The insurance & lending industries are built heavily on rule engines. Most mortgage origination in the US uses a rule engine. Just not a new technology folks discuss much.

A lot are dead– we're building a fun new one at https://www.rulebricks.com

This is the only measured take I’ve read in these comments.

If a sugar pill makes a lot of people feel better they should by every means have access to it. Within scientifically proven and “understood” mechanisms of action there are a litany of misunderstandings and MoA’s that are total guesses that folks seem so eager to dismiss in hindsight relative to homeopathy in particular. Every exception to the rule gets pulled out of the book when it comes to misrepresenting safety and efficacy for the newest juice big pharma’s pushing, like there was some sort of update to the collective consciousness rather than cheap lies and misrepresentations.

It’s silly how polarized and fixated people and western administrations are on this large quadrant of “relatively harmless, many people think it does some good”. Like it’s some low hanging fruit always there to punch down to.

Regardless, the vast majority of folks I know using homeopathy don’t actively ignore other options or deny other forms of treatment at all– if anything they respect it more and approach powerful drugs with a prudence I believe is important to preserve.

Putting aside the side of the fence I'm on for driverless cars, the steps required to operate driverless permissibly should be way more visible to the public, even if folks might not be interested in seeing/hearing any of it.

The pervasive "we're just doing it in software" and "city legislature dark arts" narratives aren't confidence inspiring.

I ride Cruise daily– I think it's entirely a scoping problem and what they decide to finalize as "FSD".

From my exp autonomous driving seems entirely plausible within large cities at slower speeds that don't instantly kill you or others, and outside that just as good a lane assist + cruise control as you can get– the rest is branding FSD and walking back a claim or two around it. It all seems very malleable.

We run a horizontal process automation business and mined GPT-3 for a bunch of industry specific use cases for our platform. I never really expected anything high-quality enough for our site, so quite surprised by how well it worked. Every piece of text on our use-cases pages is machine generated.

One cool side effect here is we're also able to run some basic tracking on which particular workflows/industries people visiting our website are most interested in.

Excited to see this on HN, congrats on the launch!

A decade ago I would've believed this to be the case

With what they've been doing in surveillance infra and hk in recent years though it does look like they're not nearly as subtle anymore

Possibly also just a change in our awareness, but I wouldn't be surprised if they write him off in a covid fatality or something