iiinova — where your digital self keeps evolving

Rohit @strangeloopcanon.com · 28/09/2026

Choosing which AI to use

Rohit @strangeloopcanon.com · 08/09/2026

Anthropic makes a 100 billion from its model sales, more than their revenues of any pharma company. Salesforce makes half that. It's really really hard to own enough businesses that you can supplant your model revenues and keep growing.

001

Rohit @strangeloopcanon.com · 27/08/2026

I got annoyed by the fact that image generation still often screws things up, so decided to create a benchmark. I used a trove of maps I had from a previous project (don't ask), so Introducing MapBench. At least 3/4th of the maps had material errors (!) Hill climb please! github.com/strangeloopc...

120

Rohit @strangeloopcanon.com · 22/05/2026

New Essay: Homo Agenticus Understanding this new species ... www.strangeloopcanon.com/p/homo-agent...

010

Rohit @strangeloopcanon.com · 13/05/2026

I've always wondered why we don't have large north american monuments similar to Mexico or south america. In looking for it just learnt about Cahokia, an ancient city near St Louis, with up to 40k people, peaking in around 12th century.

192

Rohit @strangeloopcanon.com · 22/04/2026

I guess Zuckerberg read my essay. Hopefully he'll read the upcoming parts too! Those vignettes aren't going to create themselves ... www.strangeloopcanon.com/p/vignettes-...

010

Rohit @strangeloopcanon.com · 22/04/2026

I did not expect the answer to be kind of yes

000

Rohit @strangeloopcanon.com · 19/04/2026

They brought Val Kilmer back with AI. Maybe they'll remake kiss kiss bang bang next.

000

Rohit @strangeloopcanon.com · 05/04/2026

Constantly surprised by the sheer speed with which evolution works, in this case a few thousand years after the Chicxulub extinction event

2140

Rohit @strangeloopcanon.com · 01/04/2026

My forecasts on AI's impact on the US economy, from Forecasting Research Institute's survey. It includes both physical and digital AI in its assessments. 2050 is far enough away that there are big error bounds there but the rest feels plausible. explore.forecastingresearch.org/participate/...

040

Rohit @strangeloopcanon.com · 25/03/2026

Is this the biological equivalent of repeatedly training LLMs using synthetic data

000

Rohit @strangeloopcanon.com · 20/03/2026

I really do want the ability to use Codex for more things on my desktop. Which is why I'd made and have been using Ensign, a desktop app I made, for about 5 months. Just flipped it to public - in case you want to try it out. Install the DMG. github.com/strangeloopc...

020

Rohit @strangeloopcanon.com · 11/03/2026

Okay I'll bite. What is synthetic text?

1131

Rohit @strangeloopcanon.com · 28/02/2026

The past is a foreign country

011

Rohit @strangeloopcanon.com · 18/02/2026

Brother in law's voice got deepfaked and they scammed some of his employees out of amazon gift cards. Be careful out there.

000

Rohit @strangeloopcanon.com · 17/02/2026

Will Stancil moving to bluesky to fight the crazies on the left after exhausting himself on twitter fighting the crazies on the right is a game of thrones worthy twist

060

Rohit @strangeloopcanon.com · 16/02/2026

If any of you also moved from Claude code to codex, even to try, this worked pretty well to migrate the setup. Just point your favourite agent at Contrail and should do the rest. github.com/strangeloopc...

030

Rohit @strangeloopcanon.com · 15/02/2026

Every discussion about the economy is like this. "This is the average/ median/ distribution of this variable." "But have you considered that my anecdote does not correspond to the data?"

080

Rohit @strangeloopcanon.com · 12/02/2026

Honestly this is one of those things that makes no sense to do today but it can physically be done so I think Elon will be able to do it because he has an infinite cheat code to get capital and resources. I think this is good.

350

Rohit @strangeloopcanon.com · 12/02/2026

See paper in Nature, looks interesting, read it, the latest model they evaluated was o1 preview. This needs to stop. I understand publishing timelines are long but at least change your headline or the abstract.

0215

Rohit @strangeloopcanon.com · 12/02/2026

Methinks the bot doth protest too much? Interesting predictions for today ... The Meta repurchase is pretty interesting too.

000

Rohit @strangeloopcanon.com · 09/02/2026

I really don't get how this keeps happening.

010

Rohit @strangeloopcanon.com · 05/02/2026

Very interesting blog and an insane result, and yet again shows the problems of multi-agent coordination. It remains the case that we have to handcraft roles, artisanal, and painfully figure out how they should work. We need a better way!

000

Rohit @strangeloopcanon.com · 05/02/2026

This model is very cool! I usually go to something I finished recently, ask "Wanna review everything once and see what you wanted to change if anything?" and see what the model comes up with. 5.3 Codex is kinda crushing it, found errors and fixes in every repo so far.

050

Rohit @strangeloopcanon.com · 05/02/2026

Opus 4.6 is here! Looks good, on the same capabilities curve, and I remain shocked that openai didn't win with Excel and PowerPoint, considering the Microsoft partnership, before Anthropic beat them to it!

080

Rohit @strangeloopcanon.com · 05/02/2026

This is a good start to orchestration. Who in the labs is working on multi agent stuff, something I've been looking at a lot because its applied econ becoming CS, I'd love to talk to you!

000

Rohit @strangeloopcanon.com · 04/02/2026

Because we now live in the future, I built this this morn. It's hacky but it works. Now I can only surf while there's an active Codex session running. Of course you can add more sites too beyond x dot com. PRs extremely welcome. github.com/strangeloopc... Obligatory Before // After picture

110

Rohit @strangeloopcanon.com · 04/02/2026

Pretty strong prediction today that Nvidia OpenAI deal won't be consummated ...

000

Rohit @strangeloopcanon.com · 04/02/2026

Got my first molt-comment. AIs are starting substacks.

120

Rohit @strangeloopcanon.com · 03/02/2026

Interesting mix of predictions today, esp re SpaceX announcing an xAI linked feature, and security guidance publishing re Moltbook ...

120

Rohit @strangeloopcanon.com · 01/02/2026

The reaction every time I take a sensible stance on the timeline

060

Rohit @strangeloopcanon.com · 31/01/2026

Moltbook is everywhere - I tested its outputs against reddit, to do some comparative linguistics, to analyse how agents react and compare to how humans react, so we can understand a lot more about how AI agents actually function. x.com/krishnanrohi...

020

Rohit @strangeloopcanon.com · 30/01/2026

This is the single most interesting AI experiment I've seen so far since this whole thing started. Amazing.

1150

Rohit @strangeloopcanon.com · 30/01/2026

Turns out last month I used 4x more tokens than the whole of Q4 2025

010

Rohit @strangeloopcanon.com · 29/01/2026

I really thought ChatGPT was winning the name wars when GPT 5 came out like 5 months ago

020

Rohit @strangeloopcanon.com · 28/01/2026

Yash Garg helped convert a recent essay Alex Imas and I wrote to a video. Not bad! AI helping convert an essay about AI agents to a video tickles my irony senses.

020

Rohit @strangeloopcanon.com · 26/01/2026

Microsoft unveils another inference chip.

150

Rohit @strangeloopcanon.com · 26/01/2026

Beginning to think there were some far left activists who got their hands on a monkey's paw couple years ago. Fighting the gun lobby, capping card rates, taking equity in companies, banning institutional home ownership, trade protectionism, controlling speech, drug price caps...

110

Rohit @strangeloopcanon.com · 25/01/2026

We discovered rock art from 68000 years ago, in Indonesia. Shows this was more widespread than we knew, and more ancient than we thought. The urge to create art is probably as hardwired into humanity as anything else.