Taneem Ullah Jan

Applied AI researcher and engineer

[Scott Wu @ScottWu46]

Grateful for our customers, investors, partners, and most of all for our incredible team. Hard to believe it's only been two years.

Still a long way to go to build the future of software engineering - we are hard at work every day on making Devin even better! If you haven’t tried it in a bit, check it out and let us know what you think: https://t.co/1eSXfDfReg

[Taneem Ullah Jan @taneemishere]

@Loganullyott 👋

[Taneem Ullah Jan @taneemishere]

let me be honest in this, of all the ai coding agents @cognition’s @DevinAI is most stable and snappier.

feels like a real software engineer working along side me

[Taneem Ullah Jan @taneemishere]

thank god i’m not alone at praying to screens for acting normally as before, even at openai they do this (–;

[Tibo @thsottiaux]

@mattlam_ @mweinbach We don’t. It’s all just magic. We stare at the screens and pray.

[Taneem Ullah Jan @taneemishere]

@Gerry @thsottiaux yep I have the same macos desktop app, doesn’t have the /goal

[Taneem Ullah Jan @taneemishere]

@_lamaahmad applied 🤞 hope for the best!

[Taneem Ullah Jan @taneemishere]

at this point of time, @ChatGPTapp and Codex as separate apps doesn't make that much sense tbh!

[Taneem Ullah Jan @taneemishere]

thinking inside LLM is actually thinking when you think about it!

[Christopher David @Tazerface16]

People understand that LLMs aren't actually "thinking," right?

[Taneem Ullah Jan @taneemishere]

@sandeepDshah @cognition yep, i got that

[Taneem Ullah Jan @taneemishere]

wish @cognition's devin was directly available in codex, that way it would have been a real feeling of early AGI

[Taneem Ullah Jan @taneemishere]

@theodormarcu @cognition thanks 🙏 surely will try it out!

[Taneem Ullah Jan @taneemishere]

gpt-5.5 for the win. it’s an insanely brilliant model to work with especially in codex. it feels like a far away jump than just a single +.1

[Tibo @thsottiaux]

Have you tried GPT-5.5?

[Taneem Ullah Jan @taneemishere]

even now everyone has one, but we are just distracted a bit tbh!

[Peter Welinder @npew]

Once we reach AGI, everyone will have their own neolab.

[Taneem Ullah Jan @taneemishere]

oftentimes, the (artificial) intelligence we build surprises us and makes us wonder how far we can go with this.

no better time to be alive!

[Taneem Ullah Jan @taneemishere]

@trikcode don't you guys use this strategy on your own? I've following this, use the bigger models to create the master/implementation plan, then use free/open-source and local models to execute it.

[Taneem Ullah Jan @taneemishere]

@sudoingX @arlanr 100% true, using a state of the art model for just for your few local files, doesn’t really challenge the capabilities of RAG. You’ll have to do it at scale to actually feel it.

[Taneem Ullah Jan @taneemishere]

seems like AGI won't be the AI with general intelligence; it'll be humans accompanied by several artificial intelligence systems that will be the best in their class

[Taneem Ullah Jan @taneemishere]

@_arohan_ always wondering why they use pre- and post-training but instead all these at the end of the day, are part of ‘training’

[Taneem Ullah Jan @taneemishere]

@Michael_J_Black Congratulations Michael and very best of luck for your new endeavour. You really are true inspirations for people like me. May you continue to spark motivation among us with your new position and work.