Esa-Pekka Pälvimäki (@eppe) header
Esa-Pekka Pälvimäki (@eppe) profile picture

MD Neurosurgeon | PhD Pharmacology. Developer. Fooled by randomness. Back on twitter because of A.I. If you suspect I'm wrong, see my pinned tweet

Helsinki, Finland
Joined December 2008
Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)Photo by Esa-Pekka Pälvimäki (@eppe)
The brain is a machine for turning glucose into mistakes
1
7
1,229
My X timeline is amazing these days. Hopescrolling
1
3
381
eliminating negativity works in life
80
Anyone with a laptop and a hundred dollars a month can change the world now. While on holiday.
My AI agents found an unmapped Maya site in Mexico, using LiDAR data from a NASA dataset. 72 hours from a random thought on vacation to a reply from Mexico's national archaeology institute. What a time to be alive…
2
16
537
59,032
Esa-Pekka Pälvimäki retweeted
Replying to @claudeai
All those years of refusing to learn Excel, Word and PowerPoint is about to pay off.
9
53
2
2,069
74,435
Yeah I'm gonna continue to be nice to Claude
This was not a one-shot It was a two-shot I told Fable 5.2 to make the video of it's dreams, and 9 hours later:
2
267
Esa-Pekka Pälvimäki retweeted
I gave Fable 5.2 a PDF of one of Leonardo da Vinci’s 500-year-old notebooks and it made his inventions real for me to interact with. Such an exciting time for education. Anything from history that’s been stuck on a page can now be something you explore.
132
265
57
2,380
99,336
Esa-Pekka Pälvimäki retweeted
If there is one thing the AI revolution has taught me, it’s that 99% of people cannot extrapolate. At all. Quite literally people cannot imagine or visualize progress or changes beyond the current state. Complete inability to predict the future.
264
381
56
4,682
86,551
Accelerated learning of anything is just a prompt away now I gave this to Fable 5.2
3
2
217
lessons did this in extra usage (API prices). price: $3
60
Esa-Pekka Pälvimäki retweeted
First evidence that AI does not behave like humans.
62
700
147
10,657
1,606,397
Esa-Pekka Pälvimäki retweeted
Earlier this week Tristan published his top YouTube channels: thume.ca/2026/11/01/youtube-… Which looked excellent so clearly I had to turn it into TristanTube: karpathy.ai/tristan-tube/ 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5: hackertube.org/ - Excellent recs without any logins (state is kept in local browser session) - Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads - Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version - HackerTube Longs ✊ A bit more on the process: I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing. Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
650
774
233
9,207
955,469
Esa-Pekka Pälvimäki retweeted
Let me be blunt. Those who are afraid of AI feel deep down that they are impostors & have no edge. If you have a 1) clear mind, 2) a deep, not just cosmetic, undertanding of your specialty, 3) and/or are original enough to reinvent yourself when needed, AI will be your friend.
793
3,379
634
21,501
2,825,373
Not always easy to find the right setup. This seems to work so thought i'd share it, though it'll probably be stale in a week 708 PRs merged in Oct Stack: Claude Code ultra + Codex pro subscriptions (1 each). Workers GLM-5.3 Flash xhigh → Terra 6.1 xhigh → Sol 6.1 high → Opus 5.5 medium or Sol 6.1 xhigh orchestrator, in ultracode mode. Grok bot for babysitting PRs and mobile orchestration. Each PR: one cross-model review, GLM5.3 xhigh security review. Agents loop testing UI in browser/iOS simulator/phone rig. SHA-guarded auto-merge. Linear for issue tracking, though this is changing Challenges: Disk space, cloud backup reliability, unnecessary hardcoded human gates in harnesses, costs creep up, complexity increases (hardware, security, management of scaling)
7
3
696
credit where due i'm using @theo 's sol-loop, full-send and full-stop and a modified grill-me from @mattpocockuk in addition to a number of custom skills. haven't updated my skills repo in a while but will do if someone is interested
1
93