Peter Chatwell

Founder of Pilot Generative AI and Pilot Macro Advisors. Investing in the Medway towns. Former Head of Global Macro Strategies at Mizuho.

[Institute of Economic Affairs @iealondon]

📊 On average, Brits think the NHS makes a 34% profit. It doesn't make any.

Supermarkets profits are estimated at 50%. The reality: 2–4%.

Voters think energy companies make 57%. The reality: -5% to 15%.

We are significantly overestimating how much companies profit.

[Financial Times @FT]

Labour’s lack of intellectual rigour has returned to bite it https://t.co/dVyUA15fPQ | opinion

[clem 🤗 @ClementDelangue]

Very cool to see Cursor doubling down on training great models. In my opinion, ultimately all serious companies in AI will want to train models themselves, based on open-source instead of outsourcing AI to others via APIs!

[Peter Chatwell @PeterChatwell]

@edzitron Anthropic is powered by the IP it gathers. It reverses into CRM. For CRM the poison pill is Qwen.

[Peter Chatwell @PeterChatwell]

@paulmasonnews @robfordmancs There are some other dimensions to add. All add to your argument. DM me if you want a summary.

[Alex Stauffer @alexstauffer_]

We post-trained a 3B model with RL to beat Opus on spreadsheet retrieval. Faster, cheaper, more accurate.

- If a piece of your agent loop is narrow, verifiable, and highly repeatable, a tiny trained model might beat the frontier.
- The application layer is still early and new verticals are opening fast. Cheap domain specialists orchestrated by a frontier model that only spends tokens on judgment is a bet worth watching.

[Mitchell Hashimoto @mitchellh]

Ghostty is leaving GitHub. I'm GitHub user 1299, joined Feb 2008. I've visited GitHub almost every single day for over 18 years. It's never been a question for me where I'd put my projects: always GitHub. I'm super sad to say this, but its time to go. https://t.co/DQDemHdytV

[Peter Chatwell @PeterChatwell]

@KamrynOhly @AnthropicAI @Polymarket @predictionbench We can beat you with lllama 1b in a NSAI https://t.co/08vp4D2nUi

[Peter Chatwell @PeterChatwell]

@negligible_cap @realKunalAShah https://t.co/90b9VE9mxg is what they need

[Negligible Capital @negligible_cap]

Imagine paying your lawyer $2K / hour in a high profile bankruptcy case just for them to pawn it off to slopus 4.6 and submit a key filing riddled with hallucinations

https://t.co/J3LY8zbOfQ

[Viraj Patel @VPatelFX]

⚠️ Clearest sign of UK labour slack is private sector wage growth - which fell to 2.8% YoY (lowest since Covid). Lagging the sharp fall in V/U ratio. Headline drop in UER looks noisy but in keeping with 'less bad' high-freq jobs data. This isn't data that's BoE hike worthy $GBP

[Peter Chatwell @PeterChatwell]

@ClementDelangue Definitely

[Andrej Karpathy @karpathy]

@NirDiamantAI Peter Steinberger told me that he wants PR to be "prompt request". His agents are perfectly capable of implementing most ideas, so there is no need to take your idea, expand it into a vibe coded mess using free tier ChatGPT and send that as a PR, which is now most PRs.

[Peter Chatwell @PeterChatwell]

@Teknium @cryptopunk7213 Hermes is a superb product, ahead of its time.

[Om Patel @om_patel5]

I taught Claude to talk like a caveman to use 75% less tokens.

normal claude: ~180 tokens for a web search task

caveman claude: ~45 tokens for the same task

"I executed the web search tool" = 8 tokens
caveman version: "Tool work" = 2 tokens

every single grunt swap saves 6-10 tokens. across a FULL task that's 50-100 tokens saved

why does it work? caveman claude doesn't explain itself. it does its task first. gives the result. then stops.

no "I'd be happy to help you with that." no "Let me search the web for you" no more unnecessary filler words

"result. done. me stop."

50-75% burn reduction

with usage limits getting tighter every week this might be the most practical hack out there right now

[Laura Hughes @Laura_K_Hughes]

SCOOP: A growing number of NHS staff are refusing to work on Palantir’s health data platform over ethical concerns about the controversial US tech company. https://t.co/lR05kNWVLJ

[litquidity @litcapital]

Every day is April fools when you truly believe your AI wrapper startup won’t get steamrolled by OpenAI, Anthropic, or Google

[Peter Chatwell @PeterChatwell]

@GaryMarcus Indeed, when the massive LLMs are no longer subsidising tokens, their next profit target will be modifying the truth.

[Peter Chatwell @PeterChatwell]

@FraserNelson LDI is a major factor here, they are short convexity (buy high price, sell low price)

[Laura Hughes @Laura_K_Hughes]

New: Matthew Swindells, who privately urged colleagues to add more patient data into a Palantir-built platform at the same time as he was being paid to advise the US technology company, is stepping down from his role in the health service. https://t.co/3prpwdgRNR