Tim Tairov

I build stuff.

[Tim Tairov @tim_tairov]

This is the first demo of Pill.

Pill is a small AI agent that lives on your desktop. You say what you want, or type it, and the computer does it: finding files, working across apps, carrying a task through from start to finish. It is the first project of my applied AI research lab, and at heart it's an experiment in how people and computers should talk to each other.

I've been building it for a few months and this is seven minutes of me using it as it is today, rough edges included. It runs on macOS, works with your own API key, and will be open source soon. I'm a postgrad at Durham and the lab is research done in my spare time.

I'd genuinely like to hear what you'd want it to do first.

[Rowland Manthorpe @rowlsmanthorpe]

Jensen Huang has a message for the UK

[Tim Tairov @tim_tairov]

Pill is an AI agent that lives on your desktop: you say what you want, and the computer does it.

It is the first project of my applied AI research lab, an experiment in how people and computers should talk to each other. This is six minutes of me using it, after a few months of building it.

I'm a postgrad at Durham University in the UK and the lab is open-source research, done in my spare time. Let me know what you think!

[Tim Tairov @tim_tairov]

@RiverRanx @mozo2say That looks really good!

[Tim Tairov @tim_tairov]

Your AI agents write a lot of Markdown. Here's a fast, simple way to read it.

Tim's Markdown Reader: a free, open-source macOS app. Built in Swift, with Mermaid support. Fully offline, no telemetry.

Version 1.2 out now! (Currently 11 stars and 2 issues fixed, please star on GitHub if you find it useful!) ☺️

Link to download from GitHub: https://t.co/aN2KffUbyw

[Tim Tairov @tim_tairov]

@Nguyenn259373 Thanks!

[Tim Tairov @tim_tairov]

@AwakendSnackSon If there is a lot of interest I can def make one for Linux too

[Tim Tairov @tim_tairov]

@corhoonx Cheers!

[poldish @poldidawg]

i'm ready to book so many flights

[Tim Tairov @tim_tairov]

@mozo2say Let's gooooo

[Tim Tairov @tim_tairov]

@e4kr Definitely 💯

[Tim Tairov @tim_tairov]

@XOR_AI_Hudson Thank you so much!

[Tim Tairov @tim_tairov]

@Polymarket Try working with me if you are AI and the amount of stress my ai agents go through

[Tim Tairov @tim_tairov]

@LoopandPixels Hi

[Seb Johnson @SebJohnsonUK]

Both Anthropic and OpenAI announced huge expansions plans to their London offices this year.

Anthropic is growing its office space from 200 to 800 people.

OpenAI is expanding to 500 people.

Since the OpenAI announcement it has been revealed that the London office played a huge role in the training of GPT-6 Astra.

London AI is COOKING

[Tim Tairov @tim_tairov]

@amypretzel That's such a brilliant idea! It totally solves the problem of prompting in public. I'd absolutely love to see something like that on mobile! 😊

[Tim Tairov @tim_tairov]

@charlierward Let's goooo! Congratulations 🎉

[Tim Tairov @tim_tairov]

Why I'm for AI

I build things with AI, and I've announced that I'll be organising applied AI meetups in the North East, so I wanted to explain why I'm for it. These are my opinions as of now. I don't know everything, and I'm sure some of it will change.

A lot of people worry about AI: the water and energy data centres use, the social side, AGI, and jobs being replaced. Some of that concern is fair, but I think the problems come from how people build and use the technology (including how data centres are built), rather than the creation of AI technology itself.

A single AI request uses little energy. You don't need large models for everything - you can design architectures where systems of small classifiers or micro models handle many tasks and only the hard tasks reach the energy-inefficient ones. What you build matters too, because developers make many choices: which model to use, what they will and won't build, and which servers they run on and where, since some places have cleaner power and aren't short of water. Streaming video uses a lot of energy too, and we watch it for hours without thinking twice.

To me, AI is a tool, a bit like a car. It gets you places, but it can do real harm, and we deal with that through rules and regulations, more efficient engines and electric cars, without banning cars. AI used carefully, ethically and sustainably will be useful. You can build AI that replaces people or AI that empowers them, and the question is which kind we build.

I also think people misjudge how good LLMs are. I work with them every day, and the technology is still pretty raw. It breaks often, and once the context window fills up it stops being smart.

I don't think LLMs will lead to superintelligence or AGI any time soon. They're probabilistic: they predict what's likely, not what's guaranteed to be right. Real work needs results you can trust, and with LLMs that's hard without a human in the loop. Other approaches, like world models, may get further. Most AGI talk is hype and marketing, because everyone wants to profit from it. Like smartphones, it's a big change, but not everything.

For me it's also personal. My dad was paralysed for many years, and I always used AI to double-check what the doctors told us (or didn't), because doctors in Kazakhstan are tired, have too many patients and make mistakes. Without AI I don't know what we'd have done. I come from a developing country, where good education and specialists aren't available to everyone, and I think that's where AI matters most, because it can help people become self-reliant.

We need more good people on the inside, because you don't change things by wanting to tear everything down. If people who care about ethics and the environment walk away, the technology doesn't stop. It just gets built by people who only care about profit. I'd rather there were more people in the room with good values, practical ones too, who are realistic about what it can and can't do.

[Veee @vikktorrrre]

Yann LeCun: we are not going to get human-level intelligence or AGI by just scaling LLMs.

“There’s absolutely no way in hell.”

“The idea that we’re gonna have a country of geniuses in a data center, that’s complete BS.”

“It may feel like you have a PhD sitting next to you.”

“But it’s not a PhD.”

“It’s a system with gigantic memory and retrieval ability, not one that can invent solutions to new problems.”

[Oliver Ye @_isomerism]

the moat is literally just caring about your work and being proud of your output and holy shit it does not seem to apply to 99% of people