AI Product Leader | Builder & Seasoned Operator | Translating Frontier AI into Products, Strategy & Story | ex-founder @Lilypad_Tech | prev @ibm @protocollabs
Australia
Joined July 2015
- Tweets5.6K
- Following4.3K
- Followers3.2K
- Likes26.2K
Ally Haire retweeted
Agents now make up the majority of web traffic, write more than half of new articles published online, and generate most of the code pushed to production.
This is what the early stage of a platform shift looks like: adoption is measurable, but the business models it enables mostly aren’t visible yet.
Welcome to the agent economy.
a16z.news/p/welcome-to-the-b…
Move over, B2B and B2C - the B2A economy is here
Agents are the primary users of the Internet, and soon will be the primary customers for every product
Our thesis @a16z on the massive opportunity here 👇 (cc @venturetwins)
Ive literally just been loading models I use a lot on my own hardware to reduce some costs..
Keen to try this to distribute more.
Also 100% the way in agentic engineering not just for costs but results - choosing and distributing to the right model for the job 🚀
Introducing jevgrep - a research agent CLI powered by jev from @typesafeai that reduces your coding agent cost by 40% (verified on SWE-bench)
Make sure to use the built in skill so your coding agent knows to use jg for context collection github.com/dzhng/jevgrep
4
3
185
Thinking today that it's really an incredible time to be working in tech.
Creating impact and solving complex problems has never been more accessible.
Keep building!
3
152
Im not convinced by the thesis that the Internet is becoming fundamentally more insecure.
I think, as with most things, there's a couple of things at play here:
1. We've accelerated our ability to find exploits.. which means
2. Every single code-based system is being massively tested right now - at at least 100x the speed it used to. (No idea of the real number.. lets say that's a ballpark)
Counter point though:
- We've also massively accelerated our capability to detect, block and find them first
- We've massively accelerated our ability to fix them quickly
My take is, as the agentic engineering frameworks and ecosystem matures, and the systems harden (hopefully from the inside first, not exploits), that this balances out.
Now that the Internet is increasingly becoming fundamentally insecure, including the new Linux find that if any TCP ports are exposed to the Internet it can be completely taken over, people are talking about new private internets.
The new private internet is real life. It’s clubs. It’s secret societies. It’s real physical in person connection with gatekeeping. Its tribes. It’s friend groups. It’s a room with someone at the door saying what’s the password. It’s physical files in a cabinet guarded by a dude with a gun. It’s shredding and burning sensitive documents so your security isn’t compromised. The new private internet is just what people were doing before the internet.
In a future where everything digital can be exploited, it turns out that the physical world may be the most secure of all the options.
95
Ally Haire retweeted
Yann LeCun and Randall Balestriero just Proved the Platonic Representation Hypothesis, and found a better way to achieve it.
The Platonic Representation Hypothesis suggests that AI models trained on different modalities, like images and text, eventually learn similar representations of the world. But this paper shows that how models are trained matters more than simply scaling them up.
So they mathematically show that, under certain conditions, two JEPAs trained completely separately can learn essentially the same underlying representation of the world. One learns from different images of the same scene, while the other learns from different descriptions of that scene, without ever being trained on image-text pairs.
Their approach achieves the same image-text alignment with 42x fewer training samples than next-token prediction. A 109M-parameter text JEPA even outperforms LLMs up to 72B parameters on cross-modal alignment benchmarks.
alphaxiv.org/abs/2610.39302
The timing on this post.. I created exactly this today - a fully AI generated short of me talking...
We're definitely entering the twilight zone!
I absolutely will be baking in a provenance signature to future agentic pipelines with my image
AI can copy your face and voice. One thing it can't fake 🔍
I asked @DeveloperAlly where the real opportunity is in decentralised compute.
Her answer: right where AI has problems.
"Currently, AI is a very good copycat." It can make a video that looks exactly like you and a voice that sounds exactly like you.
What it can't fake is provenance: proof of where an output came from, and that someone is who they say they are. That's a blockchain property.
Will we need proof that a video is real?
Full episode 👇
2
3
182
Ally Haire retweeted
Yann LeCun was right & nobody in this industry wants to say it out loud.
109 million parameters beat 72 billion.
He has spent years saying scale is not the thing that makes a model understand the world. Meta bet the other way. He left in November and started his own lab in Paris.
Last Thursday he published the proof.
The theory everyone has been running on since 2024 is that if you train models big enough, on text, on images, on sound, they all end up with the same picture of reality inside them. Size does it. Nobody ever proved that part.
Randall Balestriero and LeCun proved it. And the proof says size was never doing the work.
What does the work is which two things you show a model together.
A photo and the caption of that photo are two views of one thing. Models trained on pairs like that get there fast.
Two paragraphs pulled from the same document are not two views of one thing. Models trained on those get there slowly and expensively.
Next word prediction is the second kind. That is what every chatbot in the world runs on.
In their tests the first method reached the same understanding after 21.6 million examples. Next word prediction needed 900 million.
And a 109 million parameter text model ended up closer to an image model's view of the world than a 72 billion parameter language model was.
660 times smaller.
Ahh finally.. the project manager I've been dreaming of? 🙃
Definitely keen to try this out. Looks to solve a lot of my project coordination issues.
Google just open-sourced the plugin thousands of its engineers use every day to stop AI from shipping sloppy code.
It's called the Elephant-Goldfish Model.
No code until the plan survives. No AI grading its own homework. No context collapse halfway through a feature.
The idea is simple and kind of brilliant.
The Elephant is one long-running AI session that remembers everything: the architecture, the intent, every small decision. It writes zero code. It interviews you, argues with you, and keeps the record.
The Goldfish are fresh AI sessions with no memory at all. They never saw your design conversation, so they don't inherit its blind spots.
Here's how it works:
→ egm-brainstorm: you and the Elephant debate options under a strict no-code rule
→ egm-draft: you produce a Markdown design doc with the problem, the plan, the alternatives, and the exact files to change
→ egm-review: one Goldfish has to explain the design back. If it gets it wrong, the doc is ambiguous and goes back for redrafting
→ Then several Goldfish attack it as hostile reviewers hunting edge cases, hidden dependencies and logic gaps
→ A final Goldfish makes the go/no-go call: are file paths, data structures and function signatures all present and unambiguous?
Here's why this matters:
Google's team says AI made coding faster but also sped up slop, bugs and session context collapse. Speed without guardrails just scales your mistakes.
This flips the order. The design gets attacked before a single line exists.
Here's the wildest part:
This isn't a research demo. Google says thousands of Googlers run it daily to ship production code.
The repo ships the Markdown rules, the reviewer personas and the review workflows. Free, Apache 2.0.
The honest trade-off: it only runs inside Antigravity for now, with other environments coming. And security isn't on the default reviewer list. You'd add your own security Goldfish.
github.com/google/elephant-g…
138
Ally Haire retweeted
If AI does everything, what do we do? Ask the washing machine 🧺
It's the question I always try to answer, so I asked @DeveloperAlly.
Her answer: every machine that took a chore off our hands gave us time for bigger problems. "To clean up the environment, to solve disease, to solve cancer."
I've seen it in India. When the IT boom started, people moved from farms and small towns to the cities, and for the first time they had stable pay and stable jobs.
Tech is still a tool. How we use it is what matters.
What job do you think AI creates next?
Full episode 👇
"If you're the smartest person in the room, you're in the wrong room"
Lucky for me I am constantly inspired by the people, ideas, projects, innovations and novel solutions I hear daily - even on this little x-box ! ;)
I love ideas - and the people that have them :)
1
3
154
It's a rapidly evolving AI landscape... so how do you build scalable AI-native architectures that aren't outdated the moment you deploy them?
My core philosophy for working in the AI era:
🔹Code is cheap; intelligence & specification is still hard.
Models let me iterate faster than ever, but a model can only build what someone has defined well. The scarce skills are knowing exactly what to build and truly understanding what was built: AI can do the thinking, but the comprehension has to stay with us.
🔹Start from the problem, not the technology.
I build from first principles around a real user pain, often my own, and let the problem shape the solution rather than the hype cycle. A precisely understood problem is the best specification there is.
🔹Systems beat models, and composable systems outlast them.
Every company has the same frontier models. The advantage lives in the integration layer around them: the context, tools, evaluation and interfaces that make a model useful. I build that layer to be modular, so when a better model, router or standard ships, we swap a part instead of rebuilding the product.
🔹Humans own judgement; machines own repetition.
AI should take over the work no person should have to do, and put human judgement at the decisions that matter. I design agentic systems with people at the gates rather than in every step.
2
2
252
This was one of the core problems @Lilypad_Tech solved.
You dont need marketing ploys when incentives are aligned to contribute meaningfully to the network.
I remain convinced that web3 has key primitives like open markets, coordination, provenance & ofc payment rails - AI needs.
3
175
Ally Haire retweeted
It seems the days of the fully packed office are officially behind us. According to our 2026 Developer Survey, 35% of devs are coding completely remote, 36% are navigating hybrid life, and only 18% are heading into the office everyday for in-person work.
Dive into the full results on work, technology, community, and more from the Annual Developer Survey: survey.stackoverflow.co/2026…
~70% of the internet runs on php... (CRMs like Wordpress, Drupal, Magento & government & education sites).
Security firms are saying "the exploit tool successfully works against virtually every PHP release from version 5.4.0 onward"
My bet is we're going to see a lot more exploit posts like this coming out - AI has sped up both building code and the security review process.
Before anyone attacks the team... this got leaked before they could successfully patch it - clearly thats the non-ideal way to do it.
Full details of a remotely exploitable vulnerability in the core engine of *all* PHP versions will be released tomorrow, October 21st. If your organization has internet facing PHP applications then you need to be ready to patch the minute the update is released or *strongly* consider taking the application offline until you can:
php.net/index.php#2026-10-20…
288
Distributed compute is still a good bet #dePIN
📽️ Why AI Demand Is Outrunning Compute Supply youtu.be/FGC4ofTcg2k?si=Giwg… via @YouTube
Ally Haire retweeted
Streaming live from the floor 🎙️
@darrenrogan of @SuperteamAU is in the FirstLight Media pod at Solana Summit Singapore right now.
Tune in 👇
LIVE NOW 🎙️🇦🇺
Our own @darrenrogan is in the Superteam Australia FirstLight Media pod, streaming live from the floor of Solana Summit Singapore!
Pull up a chair and tune in 👇
