jmtulloss

If you are not looking at Claude usage then a plateau amongst Claude competitors in late 2025 makes perfect sense.

This is why I moved to Mercury. It’s good tech.

Stealth-ish startup right now (https://bear.build) but we’re working with some large contractors to develop the product. Assemblies are nested and go from a corner detail through walls or MEP systems to levels to the full building being the ultimate assembly.

My land is adjacent to public land and yes we have people coming through all the time. There is a forest road that becomes a private road and cuts through our land and people use it. We could put up a fence if we wanted but we don't want to... we're high enough up the hill that as long as people are being cool it's nbd. There are signs and most people (not all) are respectful.

I've found that living next to public land is an insanely undervalued luxury. There's very little I would trade it for.

I’m doing exactly this (in TypeScript) for buildings / construction. It can’t really do woodworking (not designed for that) but I can confirm that it’s an effective strategy for spatial assemblies.

Evals or it didn’t happen.

Snarky comment aside, I am very interested in how we evaluate the performance of these systems and what kinds of work match best with different approaches.

I didn't read it this way. I think I equated it most to "culture" in current companies. A culture that fosters excellence is necessary (but not sufficient) to have a great company. There are also many types of cultures that have been shown to work (ie there's not one right answer).

The article argues (and I think I agree) that how you choose to incorporate agents into your work will be a differentiator and most great companies will have a unique take on it.

With the one I’ve made, it’s best effort. We record whether or not tool calls have succeeded and if the agent dies without knowing whether or not the call succeeded, then when we resurrect it we tell it the state is unknown and it can either inspect the resource or try again depending on the side effects.

Your position appears to be based on crude production and drawing an equivalency between Middle East crude and US crude. They are unfortunately not equivalent. I agree that it’s amazing that the US doesn’t wholly depend on the Mideast for oil, but until we can process our own crude as cheaply as Mideast crude can be processed, strife in the Middle East will still push energy prices up.

Did you play with TLDraw at all? I’ve had some success with it but haven’t been using it for that particular use case.

I’m biased (my wife works there) but Fireworks or another US-based inference provider is going to be a better fit than OpenRouter. Fireworks has zero data retention by default[1] and has tools to make it easy to run your existing harness against open models [2]

OpenRouter isn’t a provider, they route to other providers, so you would need to specify which ones you’re comfortable with anyway.

[1]:https://docs.fireworks.ai/guides/security_compliance/data_ha...

[2]:https://fireworks.ai/nexus

I think the parent was referring to hosting the entire NPM registry, not having a project that uses NPM. In that case there's substantial duplication from version to version.

[disclaimer: I used to work at stripe]

I agree with you in principle but I don't think that market cap is the right metric. Market cap captures a lot of assumptions around product mix and growth rates that have been in Stripe's favor for a long time.

The most apples-to-apples comparison is just looking at payments volume and ignoring all their other products. In that case, Stripe just recently overtook PayPal! PayPal is enormous! But Stripe has that compounding growth on its side and GP's argument has become outdated in a short amount of time.

This is ignoring the bit about pc which I just disagree with.

I forked my Big Serious Harness™ that models construction projects into a harness for building a vibe coded family assistant. I couldn't figure out how to make the toy operate at toy prices until Luna. Now you can vibe code all the little apps you might want for your fam for like $5 and operate it day to day for a few cents.

I’ve been playing with this with software that needs to work with agents and also without internet at all (we’re serving construction projects that have limited access)

Obviously agent access goes away with internet failure but the state doesn’t need to… we use CRDTs and a virtual FS. There’s a toy-ish version of the harness at https://ourhearth.ai … if local first is interesting to you I’d love your feedback

This is pretty interesting but I wish it would have been refined in two ways:

- The prose before the data appears to be AI generated. Not a big deal but it makes the reader work harder to figure out what's actually being said.

- Linear didn't control for their platforms AI changes over the past year. The platform has become much more AI integrated, so a lot of numbers will move. I'm not sure how you do it but this is only useful signal with a control.

Everyone is hating but this is a hack and you love to see it.

I pushed my Home Assistant app here, you can look at it or ask your Hearth agent to copy it into your Hearth https://github.com/bearbuild/hearth-apps/tree/main/home-assi...

I just have the UI working with speakers, but the tool it has can do a ton, so you can ask Agent Heath to control all sorts of home assistant stuff.

Also, if you want the UI to be different, just ask :)

We don't have a ton of IoT providers built in right now but you _can_ hook up anything with an API by configuring a proxy with an API token. That will let the agent access the device without having access to the token.

Then you can ask the agent to write an app for you! It can just give itself tools to do stuff with the devices or can make a full UI.

What it looks like for different members of the family: everybody has access to the same stuff. Everything is syncced between all devices and any integrations you add are available to the other people in the household. Hopefully you all like each other.

I am working on this at bear.build. It's early, but I would love to get your feedback if you have some time. [email protected]

If you're only accepting crypto, you're effectively not serving the US market and the policy would be considered successful.

it appears that 3% is the threshold as 3.1% gets the nod the other way

It's because they had an AI say it instead of a human who's thinking deeply about what they're trying to say.

TBC I love AI and am totally bought in, but human taste and _severe_ editing is still required, just as it would be with any human.

Is your argument that $1500 / mo is too much? Why would the engineering team not be more rigorous in their model selection given a constraint?

Why do you think this? Confidential filings before an IPO are standard practice.

Exactly this. If there is some nuance in the article vs what Claude can tell you, then that's worthwhile. This article is just generated with a specific prompt on style but very little content editing. What's the point? It's like posting the results of a Google search. The prompt would have been more interesting.

It's not against the rules to post AI slop here, and I don't necessarily think it should be. But I do wonder how we value written content going forward. There's value to taste and style and editing and all the other human things... there's very little value in the actual words themselves. We'll figure it out.

The best stretch Github ever had was post-acquisition when Nat Friedman as CEO.

FWIW the whole group received permission to cross. The instructions were to "Truck 1 and company", not just Truck 1

The comments so far seem focused on taking a cheap shot, but as somebody working on using AI to help people with hard, long-term tasks, it's a valuable piece of writing.

- It's short and to the point

- It's actionable in the short term (make sure the tasks per session aren't too difficult) and useful for researchers in the long term

- It's informative on how these models work, informed by some of the best in the business

- It gives us a specific vector to look at, clearly defined ("coherence", or, more fun, "hot mess")