Thank you @MichaelDell. Last night, I was telling my son about your generous gift and your journey. He will definitely read your book: Play nice but win and be inspired to chart his one course of entrepreneurship.
Have been finding opportunities to extract deterministic workflows out of Agentic workflows and hard coding them. Finding much efficiency increased accuracy and improved quality
Mo Patel retweeted
If there is one constant in management as you get older it is that the companies at peak interest (by any measure) are the ones where no matter what they are doing it is a best practice and that is the source of endless fawning.
Later in a corporate arc, these same companies will do the inverse and if successful that becomes the "pivot" playbook of the day. And when not doing well whatever they were doing becomes the third rail of management.
If what you're doing works then it works. It isn't always best or the "right" answer. Context is everything. Also blindly pivoting to the other thing almost never works. You can tell from the reorg or announcement mail since it leads with the implementation and not the problem being solved (eg first, here's our new "functional" org chart, or first, here's our new "pizza team" org chart.)
Anthropic's leadership team on how they built a culture with no internal politics:
In this conversation, Anthropic's co-founders and executives explain why the company works the way it does.
It comes down to trust, low ego, and a single shared mission.
President @DanielaAmodei explains the foundation:
"I think Anthropic is really low politics... it's because of low ego. And I do think our interview process and just the type of people who work here... there's almost a like allergic reaction to politics."
That low-politics environment produces something rare: genuine trust between teams.
Co-founder @NotTomBrown describes the level of alignment inside the company:
"Everybody was just like, 'Yes, obviously we're going to do this.'"
He adds: "I do think that the trust thing is a special thing. That's extremely rare."
Where does that culture come from? Co-founder Sam McCandlish points to leadership:
"I credit Daniela with keeping the bar high. I credit the fact that we keep the reason the culture scaled."
@jackclarkSF who leads policy, notes something simple but easy to overlook:
"I think people say how nice people are here, which is actually a wildly important thing."
CEO @DarioAmodei ties it all together with the principle that keeps every department pointed in the same direction:
"Unity is so important. The idea that the product team, the research team, the trust and safety team, the go to market team, the policy team, the safety folks, they're all trying to contribute to kind of the same goal, the same mission of the company... I think it's dysfunctional when different parts of the company think they're trying to accomplish different things, think the company is about different things or think that other parts of the company are trying to undermine what they're doing."
He continues:
"The most important thing we've managed to preserve is... this idea that it's not, you know, there's some parts of the company causing damage and other parts of the company trying to repair it. But that there are different parts of the company doing different functions and they all function under a single theory of change."
General take, exceptions ignored:
This is a trap because low margin businesses cannot invest in technology. Their financial model doesn’t support it. Their investors have programmed in steady state returns. They cannot re-invest their low margin profits. This is where new entrants without the financial constraints, providing steady state returns to investors, can disrupt their businesses. You can see VCs shifting their attention to vertical AI companies that are disrupting low margin businesses for this reason. The other path is PE taking over these businesses and using debt to reinvest, reshape and repackage the business.
Mo Patel retweeted
3) Harnesses make a huge difference in cost-performance. The very simple Pi harness (@badlogicgames) got the same success rate as harnesses from the LLM vendors with Opus and GPT 5.5, but at 2x less cost! Seems to be mainly due to smaller inputs to the LLM.
Bullish on harness engineering as the product interface, for enterprise use cases the question remains general purpose or domain specific harness wins out.
new post on harness engineering for AI self-improvement: lilianweng.github.io/posts/2…
It is hard to forecast how much the future of RSI will rely on harnesses. Likely harness engineering will evolve in the direction of self-improvement and enable auto-research, and, in turn, smarter models keeps harnesses simple.
Even when many harness improvement get eventually internalized into core model, the need to specify goals and context will not disappear.
This is how AI forward companies are reshaping themselves
Since Quick launched in July 2025, we’ve collectively built more than 50,000 sites. Over 50% of employees have at least one Quick site.
When you remove the barriers to build & share, it’s incredible what happens.
More examples of how we’re using Quick :
shopify.engineering/quick
Beware AI FUD, many such cases.
I'm suspicious of that that whole story about Uber blowing their AI budget and being disappointed in the results - I dug into it and it appears to have been built on very shaky foundations
ALT I started digging into this in response to a growing volume of stories claiming that large companies were sounding the alarm because their AI usage costs had grown so large. The most widely cited of these stories appear quite overblown to me. The most discussed has been Uber, based on this report where CTO Praveen Neppalli Naga indicated that Uber had “maxed out its full year AI budget just a few months into 2026”, mostly thanks to Claude Code. Given that Claude Code only got really good in November it’s entirely unsurprising to me that a budget set in 2025 may have failed to predict demand for that tool in 2026! That Uber story was further fueled by comments made by Uber’s COO, Andrew Macdonald, on the Rapid Response podcast. I tracked down the segment and there really isn’t much there. Here’s what Andrew said:
ALT But then you sometimes go and talk to your senior engineering leaders and you’re saying, OK, how many projects that were on the cutting room floor got moved above the line because of the productivity gains because 25% of our code commits were via Claude Code last quarter? That link is not there yet, right? I think maybe implicitly there’s more that is getting shipped. But it’s very hard to draw a line between one of those stats and, OK, now we’re actually producing like 25% more useful consumer features, right? And that line is hard to draw. Somehow this fragment turned into headlines like Uber’s COO says it’s getting harder to justify the money spent on AI tokenmaxxing, because the market for stories about AI failures remains enormous.
We have seen in tech cycles where value accrues. Database companies have had the hardest time getting the combined valuations of the applications built on top of their product. As models become plug and play commodity with an optimizer sitting between the harness and model layer, value will move further up the stack.
i'm strongly against model companies focusing too much on harness, but i would love to hear if anyone has a strong argument for it
my reason against it:
if openai didn't build GPT 5.5, no one else can. this is their core competence
if openai didn't build codex cli and app, we have opencode and t3code. building harness is NOT their core competence
this is not saying products like claude code, codex aren't good - i genuinely think these are top tier products built by really talented people
my point is - the world might be a better place if model companies focus more on their core capability and give us better, faster, safer and cheaper models, rather than competing with the ecosystem in the application layer
what do you think?
8 years since first @TensorFlow Dev Summit with @JeffDean kicking things off
Mo Patel retweeted
I have Asthma, allergies, dyslexia, ADD, anxiety, and Depression.
But I will tell you that what you have does not define what you can become.
Why Not You!
Mo Patel retweeted
new post 📈 Interest in #PyTorch among researchers is growing rapidly oreilly.com/ideas/one-simple…
Mo Patel retweeted
.@Mapillary does Seamless Scene Segmentation and 3D fusion across 600m+ images
A [post + code release] with tips for:
- dealing with asymmetric graphs, variable-length images and unbalanced batches
- in-place batchnorm
- distributed training
pytorch.org/blog/mapillary-r…
Mo Patel retweeted
[v1.1.0] Official TensorBoard Support, Attributes, Dicts, Lists and User-defined types in JIT / TorchScript, Improved Distributed
Read more about the changes at github.com/pytorch/pytorch/r…
As always, get the install commands on pytorch.org
Mo Patel retweeted
We’re open sourcing PyTorch-BigGraph, a tool that makes it much faster and easier to produce graph embeddings for extremely large graphs. Quickly produce high-quality embeddings without specialized computing resources like GPUs or huge amounts of memory. ai.facebook.com/blog/open-so…
Mo Patel retweeted
Our community has a defeatist narrative even with a growing list of promptly forgotten David v Goliath stories. Isolate what matters in the long term from what gets the spotlight at this moment. You may find "irrational" courage a skill worth cultivating.
smerity.com/articles/2018/li…
