Building @RealmAlliance. ex-Palantir and other adventures.
London, England
Joined December 2010
- Tweets8.5K
- Following984
- Followers4.9K
- Likes131K
Myles Recny retweeted
New @TheCoalTrader podcast!
This week we sat down with @realmalliance CEO Eugene Marinelli to talk about bringing AI into #coal mining: belt monitoring, safety, downtime, flotation, productivity, and the very real challenge of lowering cost per ton.
Give a listen! #CoalTwitter
The Coal Boys are back this week with an interview! @mfwarder @TheCoalTrader
youtube.com/watch?v=ABJmgCIG…
Myles Recny retweeted
My first podcast interview on how we're making mines more productive at @realmalliance.
Thanks @mfwarder @TheCoalTrader for having me on!
New @TheCoalTrader podcast!
This week we sat down with @realmalliance CEO Eugene Marinelli to talk about bringing AI into #coal mining: belt monitoring, safety, downtime, flotation, productivity, and the very real challenge of lowering cost per ton.
Give a listen! #CoalTwitter
Myles Recny retweeted
compliance/legal at many small/mid caps are so terrified of litigation they are default ~6-12 month message/document retention
disproportionate amount of value will go to lazy CIOs that allow full message retention and then the distressed sale of that data. very funny
Myles Recny retweeted
A lot of people DMing me like "these results aren't REALLY that impressive, you're an idiot!" are unfortunately engaging in the very human impulse to cope. It makes sense, we've never faced this kind of thing before. But it's happening!
xitter.cf/deanwball/status/20162…
What is left for frontier humans in math and science right now?
Creative Abduction (not that kind):
"the invention of a plausible explanatory hypothesis which, if true, would make a puzzling observation unsurprising."
“our list of possible explanations is wrong because we are using the wrong entities, variables or fundamental assumptions.”
abduction requires productive hallucination
xitter.cf/Leigh_Phillips/status/…
xitter.cf/HowToPrompt__/status/2…
Google Deepmind argues that LLMs can never make real scientific discoveries.
They published a paper breaking down Albert Einstein’s private view of scientific discovery.
In a famous letter to his friend Maurice Solovine, Einstein drew a diagram of how science actually happens.
It is a cyclical loop.
First, you experience raw sensory data. Then, through a mysterious, non-logical act of intuition, you make an intuitive "jump" to abstract axioms. Finally, you use strict logical deduction to derive consequences from those axioms.
Generative AI has completely mastered two-thirds of this loop.
• Induction: Statistical pattern matching across billions of tokens.
• Deduction: Formal proof generation, like AlphaProof solving complex math Olympiads.
AI can crunch data and it can prove theorems.
But it cannot make the jump.
The paper argues that AI completely lacks Abduction, the generation of novel explanatory hypotheses when observational data is scarce.
The prevailing tech myth says that "creativity is just data compression." That if you feed an LLM enough text, scientific breakthroughs will naturally pop out.
Einstein’s formulation of General Relativity proves that is a delusion.
When Einstein formulated relativity, the observational data didn't demand a new physics framework; classical mechanics was still massively successful. The breakthrough required a conceptual rupture. An intuitive leap from physical reality to a brand-new set of foundational axioms.
An LLM can execute the math once the axioms are given. But it is structurally incapable of formulating those premises on its own.
It can interpolate inside existing human thought, but it cannot transcend it.
The translation of physical reality into formal axioms remains the absolute, hard bottleneck of artificial scientific invention.
We can build models with trillions of parameters. We can scale compute into the stratosphere.
We can make the calculator infinitely fast.
But until we solve grounding, the machine can process all the data in the universe.
It still can't make the jump.
Myles Recny retweeted
Ok dumb question, but…
Let’s assume all similar difficulty unsolved math problems fall over the next ~4 weeks
What… happens? Are their applied domains that start moving much faster? Or are we all collectively being nerd sniped?
Goodhart’s Law: "when a measure becomes a target, it ceases to be a good measure."
openai.com/index/hugging-fac…
Myles Recny retweeted
It does feel like every frontier lab is optimizing for general intelligence.
This leaves room for everyone else to optimize for task-specific intelligence. Every task requires a different point on the cost-capability frontier. Optimizing for each task involves tuning the right set of models and harness for that task to push performance and reduce cost.
I think there's plenty of room for open-weight models, optimization infrastructure (e.g. RL), vertical-specific workflows, and domain-specific models (e.g. us for document parsing) to thrive in this exponentially growing AI economy
"If you think about it like a parasite/host relationship, Chinese models are a parasite that needs to calibrate its exploitation so that the host doesn’t die."
From an ecological perspective, western frontier models are a common pool resource that Chinese open weight models exploit. They are finite, because if you exploit them too hard, they lose the incentive and resources to train new models, and everyone loses. (Chinese models aren’t magically good, they are good because Ant and OAI spend a lot of money on training).
If you think about it like a parasite/host relationship, Chinese models are a parasite that needs to calibrate its exploitation so that the host doesn’t die. Kimi arguably made a mistake. It overexploited the host and imperiled the Ant OAI growth story. So at equilibrium you’d expect open weight models to “tastefully” exploit the western frontier and choose to remain 6 months behind and slightly inferior capability wise.
However perhaps Kimi, Qwen, GLM, deepseek etc may not be playing an equilibrium game. They might be under orders to puncture the American lab growth story, even if it means killing their own host. Maybe the time is now because OAI and Ant are most vulnerable ahead of the IPO. Maybe Xi just had a bad day. Or maybe Kimi defected from the gentleman’s agreement that the Chinese labs share to not kill the host that they all rely on.
Myles Recny retweeted
If I had to restart @ElevenLabs GTM from scratch, these are the things that I would change 🖊️
- Hire quicker. I was alone selling for the first 9 months because I believed we didn't have PMF. Main reasons: the AI tourist external mantra of 2023, and the uphill battle convincing everyone to hire more GTM (Sales development, Sales, Customer Success) because we were still mostly PLG / self-service. The hiring constraint lasted until earlier this year when we boosted our team to meet demand.
- Build an enablement team earlier. Sales enablement has such an important role in an org, and we mostly think about it when we hire a lot of people. The reality is that we need Enablement earlier due to the speed of product iteration.
- Open more markets. The results from Uber v Lyft operating models played a key role in deciding to open markets early, but I would triple down on it if we were to start over.
- Have more big bets. Similar to VCs, you only need 1 big bet to work in order to generate Alpha. We could have executed more big bets if we had had a single person owning & iterating on them. Former successful founder profile with drive and speed.
- Be more vocal about product ROI & our success. People buy from people, and companies buy based on success & ROI. In AI, we've all been slow showing ROI. This should be the key number when you start a conversation.
- Hire senior sellers. In startups, we've been told that culture is everything and someone with +25y experience may struggle to fit in. This is bullshit. The experience, connections and drive +25y sellers bring is crazy. We have a bunch today, but I would hire many more earlier.
- Reduce cash commissions and replace them with more equity. Cash is king, but Equity is queen. Queens also rule countries, so why not make Equity the dominant one in GTM?
- Open more offices. We are a remote-first company, but our customers are everywhere, and employees want to meet each other. I would open more offices to be embedded in markets quicker.
- Fun brand. Brand perception is 33% of the work. Building a fun brand adds quickly.
- Outbounds outbounds outbounds. We waited too long to migrate to an Outbound organisation. I would do it from day 1 now.
- Pipeline construction with more whales. The ideal pipeline needs to have a combo of segments to reduce risk, improve results and provide liquidity. We waited too long to pitch & bring whales.
- Not all deals are good deals. We, sellers, want to close a lot of ARR, but it is important to split early on between good vs bad deals.
Thank you to Tomasz Gmur for asking how I would do GTM again!
"The Applied AI layer has a huge opportunity to combine frontier intelligence with open or cheap closed models to orchestrate workflows in any given domain. Due to evals, deep domain context, being trusted with enterprise data and workflows, this layer can maximize performance and cost combination. The applied AI layer will also often have their own RLed models especially for high volume, predictable tasks in their systems."
A few thoughts on what we will see in AI structurally for the foreseeable future:
* Frontier intelligence continues unabated and pushes the industry forward continuously. The top labs will continue to buy the best and the most data, build the most compute, be at the forefront of improved training breakthroughs, and so on. A few different approaches stratify the market on pricing and capability, but overall competitive pressure brings down pricing on a per task basis. That said, we just ask more from the models over time - as one thing gets cheaper, we just use more - so frontier spend and use remains robust.
* Open weights rapidly absorbs frontier breakthroughs (and drives other breakthrough directions given the constraints), offering both lower cost intelligence and the ability to be post trained for specific workflows and domains. This creates a healthy counter balance to the frontier as you can run models “at cost” on a hyperscaler at any time, and tune models just for your tasks.
* The Applied AI layer has a huge opportunity to combine frontier intelligence with open or cheap closed models to orchestrate workflows in any given domain. Due to evals, deep domain context, being trusted with enterprise data and workflows, this layer can maximize performance and cost combination. The applied AI layer will also often have their own RLed models especially for high volume, predictable tasks in their systems.
* Individual enterprises will generally focus on their enterprise context, making sure they can get any AI system the right data and information to work with, in a continuously improving way. Some will go off and train their own models for specific areas of work (large banks, pharma, etc.) where they can get real alpha from doing so given the many tradeoffs, but most will spend energy on making sure they can get all of the gains from AI breakthroughs on their data and workflows.
Net net: even though some of this gets framed as zero sum, there’s just a ton of opportunity for all layers of the stack and approaches.
Myles Recny retweeted
I've heard a lot of questions about Fable's availability on subscription plans.
While it will come off subscriptions after July 7th, we aim to restore Fable as a standard part of our subscriptions as soon as capacity allows, as we mentioned in our original blog post.
Myles Recny retweeted
As engineering, product, design, DS, etc. melt into a new kind of role, I was reflecting on what roles might look like in the future. For example, when I look at the Claude Code team I see what I think is five archetypes:
1. Prototyper: comes up with brand new ideas; churns out many ideas, most of which don't ship
2. Builder: quickly turns a prototype/idea into production-grade product/infra
3. Sweeper: cleans up the UI, simplifies the code and system, unships, optimizes performance
4. Grower: takes a product that has been built and iterates on it to improve Product-Market Fit
5. Maintainer: owns a mature system to make it secure, reliable, fast, and efficient as it scales
Many people span across 2 roles, and sometimes 3 roles. I also notice that these roles are not really tied to job function -- eg. across Anthropic, some designers match category 1, some 2, some 3; same for engineers, PM, DS.
A healthy team needs a mix of these, depending on the product:
- A product that is new and pre-PMF needs people that are strong at 1+2+3
- A product that is growing and has found PMF needs 2+3+4 and some 5
- A product that has strong PMF needs 3+4+5 and some 2
Maybe product roles of the future will look more like this, and less like the domain-specific roles of today?
Myles Recny retweeted
Instead of “don’t look at the code at all” or “read every line in detail”, you can explicitly instruct the LLM to write code that’s as easy as possible to read
Myles Recny retweeted
A lot of alpha in the simple prompt: "Optimize the speed of <any slow thing>. Keep iterating to make it faster and faster."
Myles Recny retweeted
The job of a software engineer is becoming more like software auditor — someone who certifies the code and is held accountable for it
