Dan Cobley

Isometric has raised $40 million in Series A financing from AVP, Plural, Lowercarbon Capital, and legendary venture investors John Doerr and Walter Kortschak. Heavy industry is back. Across manufacturing, power generation, data centers, waste management, and more—the industrial economy is seeing a resurgence. But every industrial project needs to be certified against regulatory, safety, and sustainability standards. Until now, certification involved a tradeoff: Do it fast, or do it right. Never both. Isometric is the world’s first AI-native certification company, deploying AI agents to make the certification process instant and invisible, unleashing the potential of the industrial economy. We proved it in carbon removal—where we’re now the largest certifier globally, contracted across 200+ projects by the likes of Microsoft, Anglo American, and Boeing—and are doing the same for energy, fuel, and building materials. Now we’re bringing agentic certification to the rest of the industrial economy. Richard Tyler told our story in The Times: https://lnkd.in/e_QxNsKm

We’re hiring a Growth Lead. At Early we’re building a new way to detect serious disease far earlier by making it super easy from anyone’s home. No needles. No expensive clinics. No sending samples to a lab. We bring the lab into your home with powerful science that is magically simple to use. We’re looking for someone exceptional to help us take our mission to millions of people. If you have the experience and character to lead Growth at Early then message me directly. If you’re the right person then you will know what we are looking for. Please note we are not a remote working or hybrid company. We are bringing very advanced technology into people’s homes but we can’t build this from our own ones. Please don’t apply if you want to work from yours. #growth #healthtech

What happens when one marketer can do the work of ten? At Cannes Lions next week, I’m excited to host a group of marketing leaders for a conversation on how agentic AI is reshaping the future of marketing teams. We’ll explore how AI is creating a new generation of marketers with unprecedented leverage and what that means for hiring, team structure, and leadership. We’ll discuss: • What 10x marketing output actually looks like in practice • How the marketer’s role is evolving with agentic AI • How leaders can build AI-native marketing organizations Looking forward to a candid discussion about what’s changing, what’s working, and what’s coming next. If you’re at Cannes and interested in joining, DM me or respond at the link in the comments.

The latest edition of our Board Value Index landed in the Financial Times today. Here's what the data shows. We surveyed more than 400 directors, CEOs, and CFOs across the UK, US, Nordics, and Middle East. Globally, 98% of directors say their board has taken up the question of AI in decision-making. Yet only 49% are actively reviewing it and moving from discussion to action. That gap matters because while boards are debating AI, many are still struggling with the fundamentals. 84% of directors say poor information quality has caused at least one delayed, rushed, or poor decision in the past six months. And only 37% of directors globally see their board as an essential tool for value creation. The full findings are in the first comment.

We're pleased to share a new report, in partnership with Management Today, on board effectiveness. Only 33% of directors view their board as an essential tool for value creation. Around half say their board could perform better. These findings, from our Board Value Index, are evidence of a persistent gap between board potential and performance. Our latest report takes a data-driven look at why this gap persists despite record investment in governance. It reveals what's structurally getting in the way, what high-performing governance teams do differently, and how AI is changing the equation for governance professionals. Download the full report, Boards on autopilot: Why modern governance teams need to embrace AI… now. Link in the comments.

How good are you at ‘tokenminning’? Recently, there’s been a lot of focus on ‘tokenmaxxing’, where companies are encouraging people to use lots of AI, and people are bragging about how many AI tokens they are paying for as a result. There have been stories of engineers using 200 billion tokens in a week, with some reportedly designing deliberately inefficient tasks so that AI wastes a huge amount of tokens – and also helps them top the internal ‘tokenmaxxing leaderboard’. It’s a trend that is unlikely to last. As well as the misplaced incentive of encouraging employees to maximise cost rather than deliver value, there’s the problem that AI is currently heavily subsidised by providers. Teams are therefore starting to get nasty surprises as they realise their workflows rely on models that are about to get more expensive, or less reliable, or both. We may be about to see a summer of ‘tokenminning’, as teams scramble to contain unpredictable, exploding costs. To help you get ready for the slender times ahead, here’s a light hearted weekend challenge. Below are a series of tasks. Your job is to communicate the task in as few words as possible, in a way that means a popular AI tool (e.g. ChatGPT Auto) will output the right answer when given that prompt. For example, if the task is “Who is the Prime Minister of the United Kingdom?” Then the prompt “UK PM” will generally get the right answer out of the model despite using only 2 words/tokens of input text. (Note: “UKPM” would also count as two words, because AI will split it into multiple tokens when it processes text. You can explore more here: https://lnkd.in/eDRSwJMK) Below are six more tasks for you to try. I’ll post some solutions as a comment tomorrow. And do comment below if you manage to find a solution with even fewer words than the limit! 1. When was Shakespeare born and when did he die? (2 word limit) 2. What elements are in the top row of the periodic table? (2 word limit) 3. Give me Python, JavaScript and C++ code for generating a random number (3 word limit) The tasks above focus on minimising input words/tokens. But generally output tokens are more expensive that inputs, so there can be a benefit to using more tokens in the instructions to reduce the amount of work a model does in generating an output. For example, “UK PM name only” uses two more input words, but typically shortens the (more expensive) output by over a dozen words. Here are three more tasks, where you have to minimise both input AND the output. 4. How old was Shakespeare when he wrote Romeo & Juliet? Output only a single number with his age. (6 word input limit) 5. What is the chemical formula for water? Output only the formula itself (3 word limit) 6. Give me a Python function that prints “hello world”. Output only the function. (4 word limit) I’ll post some solutions as a comment tomorrow. And do comment below if you manage to find a solution with even fewer words than the limit!

We're hiring a Junior ML Engineer in London Unbox processes millions of TikTok videos a day through frontier model APIs. The output feeds brand intelligence tools used by some of the world's largest consumer brands. The role: writing Python services that orchestrate LLM calls at scale, working alongside our senior ML engineer to take research into production. This is an early-career role. We care more about how you think than how long you've been doing this. If you've written Python beyond coursework and you're curious about LLMs in production, apply. Full details here: https://lnkd.in/d7qMR_ih #hiring #mlengineer #london

For decades, gravity has dictated how drugs are made. Today, we are changing the game. BioOrbit has closed a £9.8M ($13.2M) round, the largest seed round in the history of in-space manufacturing. But this isn't about "space tech." This is about the £1 trillion global antibody market. Led by Cambridge scientist Katie King, PhD and Oncology Researcher, Dr. Leonor Teles, BioOrbit is doing what was once considered science fiction: moving the factory into orbit. By removing gravity from the equation, we are turning complex cancer therapies into ultra-high-concentration, fluid formulations, injectable at home and always accessible. The Result - Cancer treatment is moving from the cold sterility of a clinical ward to the sanctuary of the home. The Impact - A 90% reduction in hospital drug administration costs and a total paradigm shift for the $1 trillion antibody market. This round, co-led by LocalGlobe and BREEGA, with support from Seedcamp, Auxxo Female Catalyst Fund, Type One Ventures, 7percent Ventures and others, brings more than capital. It represents a global trust in a new industrial era. It is the moment a female-led deep-tech powerhouse proved that the most difficult problems on our planet are best solved 400km above it. The future is here. #BioOrbit #SpaceManufacturing #DeepTech #CancerResearch #HistoricSeed #WomenInSTEM #FutureOfPharma #SpaceEconomy

Only 23% of directors believe their board is operating at its full potential. Nearly a third say their board adds no value at all. That is what we found in our own Board Value Index, and what our CEO Pippa Begg spoke about on CNBC Squawk Box Europe earlier this week. The problem is not talent. Boards are under-resourced, stuck reviewing backwards-looking information, and rarely getting enough on strategy or the external environment to make the decisions that matter most. Our AI sparring partner changes that. It is built on 20 years of boardroom expertise, optimised across 130-plus topics and 80-plus skills and capabilities, with secure hosting, data retention in your control, no machine learning on your data, and strict permissions and information rights throughout. Boards are capable of far more than most are delivering. We are here to close that gap. See how we are building AI for the boardroom: https://bit.ly/4vQEdOA