Gotta love that Github has decided that hiding largest diffs by default was a good idea. It's not like the amount of changes has any correlation to importance, right? And to top off the joke that they are, it's still amazingly slow, even when you skip the bulk of the code.
At this point I don't know if anyone at github does code reviews. They hide large diffs as you said, but navigation is also broken, I have to click several times on a filename for the UI to scroll to it.
They also decided that comments should not be visible instead you have to hunt for a tiny avatar of the person who left them. Very puzzling
At least they finally fixed the sidebar to use anchor tags instead of divs, so right-click can open the file in a new tab. It's still a bit wonky, since the a tag is only the size of the filename, with zero padding. It was broken since they introduced their new UI sometime last year.
Maybe by the end of 2027 clicking a filename actually takes you to it.
> @0xTMK: Why do companies acquire open source software companies if everyone can already access the repos?
> @mitchellh: Because code by itself isn't valuable. Coding is the easy part.
So... acqui-hire? No hate on that, but this team must be wildly talented.
That question isn't really applicable to this situation, since https://code.storage is the company's main product. It has a business model, customers, and revenue. The open source libraries are tangential.
if code by itself isnt valuable, why are companies reportedly trying to increase their code output with ai while letting people go or letting their skills rot? which is it, is code valuable or isnt it?
We need a term for an acquisition where you buy an entire company because it would be cheaper than paying a monthly subscription fee to use their product.
Whenever I see this acquires or hiring of successful products founders or teams by big labs I ask myself if, apart for the usual obvious reasons that any non ai model company would want them, they are also reusing this human talent to build the dataset of "how the sausage is made".
>Code.Storage powers the largest agentic workloads in the industry to store code and other machine generated artifacts at scale using git.
>Lovable, Poke, Anything, Bolt, Amp, Parahelp, and many others rely on Code.Storage to store their generated code, memory files, design artifacts, legal documents, and other product files and resources that change over time.
I'm interested in knowing more about the company itself. How many people is/was it? Were they working all in an office together or local-remote, or fully globally distributed? Was the focus massive scale, or ultra-low latency (for agent iteration), or agent-first (but human in the loop) usecases that could be enabled by having core control of a git service and the wrappers around it?
Why not just visit the company's website? It would take less effort than writing this comment. Why is HN a hyperbolic bikeshedding chamber where everyone is trying to outnitpick each other?
Unfortunately, their website is also not very helpful.
You'll have to click the link through to https://code.storage to understand: They host Git servers at scale, especially for agentic companies like Lovable (who create many projects and need many code repos to store them).
Site currently seems to be broken, as it eventually just returns a "Dang" error message with a picture of a tombstone. How quirky and cute! Just like the insufferable po-mo marketing on the twitter account.
Words have meaning. I really hate PR that just has no respect for language. To me that reads like "Toxic waste, lovingly dumped into the river with a dump truck."
When an HN link goes to a corporate PR or marketing landing page, I give it about 15 seconds to explain itself and then I just bail to the comments - which is based on the best marketing tip I ever got.
As a coder turned tech startup entrepreneur a few decades ago, I had to learn marketing, advertising, PR and branding from scratch. Fortunately, I had some kind early mentors who were old-school agency types with tons of experience. The single best nugget was from a grizzled Mad Men-esque pro who made it simple. Put 3 things into the first ~3 sentences: 1. A clear, yet mom-level, "What it is", 2. a more advanced "Why it matters", 3. At least one thing that's new, different and important - which is NOT true for the three closest competitors (or our last version). The pass/fail for any ad, landing page or PR was showing drafts to different strangers and testing if they got those 3 things in the first ~15 seconds. If they didn't he'd say, "They're not dumb - your marketing is dumb."
When I complained it was hard, if not impossible, to do that. He dropped the killer line: "Then you don't have a marketing problem. You got a product problem." Over sandwiches he clarified, "You're the genius... just pick problems and build your thing so it makes those three sentences easy to write." He basically thumbnailed a master class in category selection, positioning and differentiation in one lunch. Solid gold - which I've shared with dozens of people in the decades since.
One of the primary reasons HTML supports links is to avoid always having to repeat information that's already available elsewhere.
The first sentence of this submission includes a link to the company's homepage, which states what the company does and what its products are. There's no need for every announcement to be entirely self-contained when the relevant information is just a click away.
As someone with a cat who used to be named "Vito", but eventually ended up as "Pierre", I'm going to assume "Pierre Computer" somehow makes your AI act more French. :P
OpenAI lacks process knowledge and technical experience, so they acquire it. They are so lacking that acquiring almost any established tech company needs no justification.
When your company is bleeding billions every year already, buying startups is pocket money. If anything it's actually a good diversification strategy to build tools that bring new value not directly tied to their loss leader product.
I mean good for the team, I'm sure they will make out quite nicely.
But it was really great to see high quality OSS being extracted from their enterprise-focused https://code.storage/ service. Not I imagine the oss updates and new libs will decline, maybe not right away, but eventually, especially as they get rolled into whatever the product-offering of the week is that altman cooks up.
OpenAI continues to do everything right, using their position to acquire as many companies as possible. If you don't take a peak at the horror show that is Anthropic everything would look good.
Musk envy. Cursor was also building large scale git infrastructure. Pierre Computing does the same.
All these companies need success stories but only end up making dev tools, a notoriously unprofitable market. The corporate market is already saturated with LLMs/agents because of the dictatorial forced rollout, so there is actually no growth in sight at all.
I'm thinking herd mentality because software development is where people actually pay for AI. Not that the Cursor acquisition will ever pay for itself.
Funny how small business is resembling product at a slave market, where they are trying to show nice teeth and muscles, hoping that AI master or another whale will buy them.
At least it is a signal AI is not good enough yet, so that they could clone it in one prompt.
I have to imagine this is related to: https://openai.com/index/scaling-git-at-openai/
And you'll see that mitchellh is a big fan of Pierre (and I think investor)
https://x.com/mitchellh/status/2036855931569651884?lang=en
For those who don't know Pierce, here's a demo from mitchellh:
https://x.com/mitchellh/status/2057229385963618787
Gotta love that Github has decided that hiding largest diffs by default was a good idea. It's not like the amount of changes has any correlation to importance, right? And to top off the joke that they are, it's still amazingly slow, even when you skip the bulk of the code.
At this point I don't know if anyone at github does code reviews. They hide large diffs as you said, but navigation is also broken, I have to click several times on a filename for the UI to scroll to it.
They also decided that comments should not be visible instead you have to hunt for a tiny avatar of the person who left them. Very puzzling
Yeah, unlikely they use Github much internally.
At least they finally fixed the sidebar to use anchor tags instead of divs, so right-click can open the file in a new tab. It's still a bit wonky, since the a tag is only the size of the filename, with zero padding. It was broken since they introduced their new UI sometime last year.
Maybe by the end of 2027 clicking a filename actually takes you to it.
He is a minor investor. He commented on the acquisition at https://x.com/mitchellh/status/2118012675582280038
Who needs to review millions lines diff ? Even 10k lines diff ? What can you do with that ?
tangent: I think mitchellh was talking about moving Ghostty off GitHub? Any update?
i found Mitchell's answer to someone's question regarding why buy a open source software company, very interesting.
https://x.com/mitchellh/status/2118017893921436019?a=bCZdMwC...
code/coding isn't valuable. the problem solving skill is what set a good and bad engineer apart.
For those too lazy to open the Twitter/X link:
So... acqui-hire? No hate on that, but this team must be wildly talented.That question isn't really applicable to this situation, since https://code.storage is the company's main product. It has a business model, customers, and revenue. The open source libraries are tangential.
if code by itself isnt valuable, why are companies reportedly trying to increase their code output with ai while letting people go or letting their skills rot? which is it, is code valuable or isnt it?
We need a term for an acquisition where you buy an entire company because it would be cheaper than paying a monthly subscription fee to use their product.
Acquiscription, n. - The acquisition of an entire company after reviewing its enterprise pricing.
brilliant
Buy-to-Unsubscribe
Subscription avoidant acquisition strategy
why would owner sale company at loss?..
SubSkip acquisition
I read good things about their diffs library: https://diffs.com/
Whenever I see this acquires or hiring of successful products founders or teams by big labs I ask myself if, apart for the usual obvious reasons that any non ai model company would want them, they are also reusing this human talent to build the dataset of "how the sausage is made".
Also the file tree library: https://trees.software/
Sadly this level of attention to detail is going to extinct
I am actually using it in the code forge I am building, https://juju.bi
In fact, Pierre's earlier iteration was a code forge and mine is heavily inspired by it.
Wow, that page crashed my iPhone!
Cool. Pierre Computer is a company founded by original Bootstrap devs: Jacob Thornton and Mark Otto
Correction: Only Jacob Thornton is one of the co-founders of Pierre; the other one is Ian Ownbey (sources: https://presentofai.com/spotlight/pierre.co, https://cellcog.ai/blog/openai-acquires-pierre/)
>Git infrastructure at scale
>Code.Storage powers the largest agentic workloads in the industry to store code and other machine generated artifacts at scale using git.
>Lovable, Poke, Anything, Bolt, Amp, Parahelp, and many others rely on Code.Storage to store their generated code, memory files, design artifacts, legal documents, and other product files and resources that change over time.
AI bros buying AI bros. All those companies in the list have ca. 10x inflated valuation.
I bet in 5 years nobody will remember 95% of them.
AI bros or not, presumably they have better uptime than Github
Because nobody is using them... My service has 99.99999 uptime btw...and zero users.
I'm not sure that's fair. Look at their customers -
https://code.storage/changelog/lovable-case-study
I'm not sure if it's fair to compare GH's uptime vs a barely known "Pierre Computer".
https://xnotme.com/pierrecomputer/status/2117999036175188183
I'm interested in knowing more about the company itself. How many people is/was it? Were they working all in an office together or local-remote, or fully globally distributed? Was the focus massive scale, or ultra-low latency (for agent iteration), or agent-first (but human in the loop) usecases that could be enabled by having core control of a git service and the wrappers around it?
Well Claude is French and Pierre too so....
Claude is American both in the company and in the namesake. He was born in northern Michigan.
https://www.etymonline.com/word/Claude
masc. proper name, from French form of Claudius.
Claude Shannon
You mean Jean Claude...Jean-Claude Van Damme.... https://youtu.be/M7FIvfx5J10
You’re not a very good troll huh
Also not French.
French speaking part of Belgium. The language is the most relevant factor for what is being talked about.
I wish these "acquired by" announcements started with an explanation of what the startup was in the first place.
I had never heard of them before and have no clue what they do or why OpenAI wanted to buy them.
Why not just visit the company's website? It would take less effort than writing this comment. Why is HN a hyperbolic bikeshedding chamber where everyone is trying to outnitpick each other?
Unfortunately, their website is also not very helpful.
You'll have to click the link through to https://code.storage to understand: They host Git servers at scale, especially for agentic companies like Lovable (who create many projects and need many code repos to store them).
Why context when discuss???
Site currently seems to be broken, as it eventually just returns a "Dang" error message with a picture of a tombstone. How quirky and cute! Just like the insufferable po-mo marketing on the twitter account.
Isn't that literally what's in the first paragraph of the announcement?
https://openai.com/index/openai-to-acquire-pierre/
It is, but the HN post does not link to that.
> Craftsmanship meets scale for the agentic era
Words have meaning. I really hate PR that just has no respect for language. To me that reads like "Toxic waste, lovingly dumped into the river with a dump truck."
When an HN link goes to a corporate PR or marketing landing page, I give it about 15 seconds to explain itself and then I just bail to the comments - which is based on the best marketing tip I ever got.
As a coder turned tech startup entrepreneur a few decades ago, I had to learn marketing, advertising, PR and branding from scratch. Fortunately, I had some kind early mentors who were old-school agency types with tons of experience. The single best nugget was from a grizzled Mad Men-esque pro who made it simple. Put 3 things into the first ~3 sentences: 1. A clear, yet mom-level, "What it is", 2. a more advanced "Why it matters", 3. At least one thing that's new, different and important - which is NOT true for the three closest competitors (or our last version). The pass/fail for any ad, landing page or PR was showing drafts to different strangers and testing if they got those 3 things in the first ~15 seconds. If they didn't he'd say, "They're not dumb - your marketing is dumb."
When I complained it was hard, if not impossible, to do that. He dropped the killer line: "Then you don't have a marketing problem. You got a product problem." Over sandwiches he clarified, "You're the genius... just pick problems and build your thing so it makes those three sentences easy to write." He basically thumbnailed a master class in category selection, positioning and differentiation in one lunch. Solid gold - which I've shared with dozens of people in the decades since.
Commenting also to remember... as well in case someone says "HN is dead!"
HN has a favoriting feature.
Commenting to rmemebr this but it’s good
One of the primary reasons HTML supports links is to avoid always having to repeat information that's already available elsewhere.
The first sentence of this submission includes a link to the company's homepage, which states what the company does and what its products are. There's no need for every announcement to be entirely self-contained when the relevant information is just a click away.
As someone with a cat who used to be named "Vito", but eventually ended up as "Pierre", I'm going to assume "Pierre Computer" somehow makes your AI act more French. :P
The AIs are now on strike.
OpenAI lacks process knowledge and technical experience, so they acquire it. They are so lacking that acquiring almost any established tech company needs no justification.
or how much they got bought for...
https://openai.com/index/openai-to-acquire-pierre/
https://github.com/kohoj/codex-app-teardown they were using Pierre libs for a while, so only natural.
I like the name "Pierre Computer Company".
Me too.
"Hmmm-mmm, I smell pork chops. Does your dad cook?"
Remember when OpenAI bought a device company for 6 billion dollars, a company that had never made a device.
Remember when OpenAI bought a podcast that gets 500 views and no comments on Youtube for 100 million dollars?
(And they've never advertised on it either, their name isnt anywhere on the screen while theirs dozens of other sponsors)
Im convinced Sam is just passing money to his friends, bribing or something idk. Something seems off.
It’s the circular economy we keep hearing about, investors put money into OpenAI and they send it to silly, barely profitable projects.
When your company is bleeding billions every year already, buying startups is pocket money. If anything it's actually a good diversification strategy to build tools that bring new value not directly tied to their loss leader product.
Huh?
No.
Neither of these companies made money, the podcast nobody watches literally made more money than the 6 billion dollar device company.
You dont buy companies that make negative amounts of money and are newer than you. No idea where you got that idea.
> You dont buy companies that make negative amounts of money and are newer than you.
Someone go back in time and warn Zuckerberg not to buy Instagram
TIL the founders of the company are not all named Pierre :(
nooooo.
I mean good for the team, I'm sure they will make out quite nicely.
But it was really great to see high quality OSS being extracted from their enterprise-focused https://code.storage/ service. Not I imagine the oss updates and new libs will decline, maybe not right away, but eventually, especially as they get rolled into whatever the product-offering of the week is that altman cooks up.
Oh well. so it goes
OpenAI continues to do everything right, using their position to acquire as many companies as possible. If you don't take a peak at the horror show that is Anthropic everything would look good.
what else is left out there to build and get bought
Refreshing, compared to "our journey" and "our mission" slops
The plural of slop is slop.
Who? https://pierre.computer/
Musk envy. Cursor was also building large scale git infrastructure. Pierre Computing does the same.
All these companies need success stories but only end up making dev tools, a notoriously unprofitable market. The corporate market is already saturated with LLMs/agents because of the dictatorial forced rollout, so there is actually no growth in sight at all.
It's not Musk-envy, it's just everyone is running into similar scaling issues at the same time. Agentic coding is breaking Github, so everyone is moving towards new ways of working with git. Anthropic built their own system https://www.anthropic.com/research/redesigning-source-contro... and so did OpenAI https://openai.com/index/scaling-git-at-openai/
you think OpenAI needs success stories and is therefore pivoting to dev tools?
I'm thinking herd mentality because software development is where people actually pay for AI. Not that the Cursor acquisition will ever pay for itself.
Yes ELIZA, OpenAI needs success stories. As does Anthropic. This zero growth LLM coding scam will collapse at some point.
Lot of new accounts like this created. Looks like warfare out there.
Funny how small business is resembling product at a slave market, where they are trying to show nice teeth and muscles, hoping that AI master or another whale will buy them.
At least it is a signal AI is not good enough yet, so that they could clone it in one prompt.