Taymour Khan

Co-founder of Corso

[Taymour Khan @taymour]

I built Corso Labs: ED Documentation Review Room for the Gemma 4 hackathon.

It uses Gemma 4 31B on @Cerebras to run an 8-agent, multimodal ED documentation QA workflow over a synthetic raster medical-record image.

The demo reads facts from pixels, runs a visible agent chain, returns a clinician-editable physician note review packet, and compares Cerebras speed against a GPU path.

The point: speed changes the product shape.

On slow inference, multi-agent review feels batch-like. On Cerebras, it can feel interactive.

Live app: https://t.co/tlSoOuZXXZ
Code: https://t.co/GnwbWAmKV2

Synthetic data only. Not diagnosis, treatment, clinical validation, or production software.

@googlegemma @GoogleDeepMind @cerebras

[Taymour Khan @taymour]

@NousResearch @_weidai THE KINGS!

[Taymour Khan @taymour]

@iamsaranhere @thsottiaux has blessed me

[Taymour Khan @taymour]

i dont sleep until @thsottiaux sleeps. Waiting for 2a CST to begin T_T

[Taymour Khan @taymour]

@LLMJunky very cool tip!

[Taymour Khan @taymour]

i just genuinely think the interface and UX for GPTwork is not well integrated. i think even the previous UI was better because you could integrated GPT chats better with Codex. Work just feels now like the only way to access GPTChats and a gimped version of the Codex tab (bc even skills doesnt share over). It genuinely feels like a useless tool thats only for non-techies

i know that sounds super critical, i love OAI, Tibo, 5.6, everything, i just think this has a lot of needed improvements. I have so many ideas T_T I shared them in another tibo post. A lot of people have some great ideas as well

[Taymour Khan @taymour]

@thsottiaux https://t.co/b0gZ3xkMhX

[Taymour Khan @taymour]

- the new bidi voice thing for codex
- syncing features between gpt(work)/codex (like SKILLS)
- making code spans double clickable so it only highlights that and not the whole line
- incorporate more FINANCIALS into Codex, currently stuck on GPT only

also much easier way to use the chats from gpot on the codex side -- they seem too segregated and it's difficult to "add to chat" like you used to. I have to "@" tag the thread in the codex textbox. It seems more difficult to leverage imo

[Taymour Khan @taymour]

@thsottiaux id love to be a part of this -- im trying to see how i can best integrate both in my life and see where certain features overlap or are best done on work vs codex. There's a lot of cool features i think could be added and im always tinkering

[Taymour Khan @taymour]

- the new bidi voice thing for codex
- syncing features between gpt(work)/codex (like SKILLS)
- making code spans double clickable so it only highlights that and not the whole line
- incorporate more FINANCIALS into Codex, currently stuck on GPT only

also much easier way to use the chats from gpot on the codex side -- they seem too segregated and it's difficult to "add to chat" like you used to. I have to "@" tag the thread in the codex textbox. It seems more difficult to leverage imo

[Taymour Khan @taymour]

@ravikiran_dev7 "just"

[Taymour Khan @taymour]

@thsottiaux @Ananth7e setup an MCP to the reset button, then consider open sourcing it

[Taymour Khan @taymour]

@thsottiaux also much easier way to use the chats from gpot on the codex side -- they seem too segregated and it's difficult to "add to chat" like you used to. I have to "@" tag the thread in the codex textbox. It seems more difficult to leverage imo

[Taymour Khan @taymour]

@PorgimusPrime @thsottiaux not discrediting what ur saying or anything -- but i swapped from opus to gpt bc opus spoke in long soliloquies, it was unbelievably frustrating to sit there and read sagas. GPT has been much better imo

[Taymour Khan @taymour]

@thsottiaux - the new bidi voice thing for codex
- syncing features between gpt(work)/codex (like SKILLS)
- making code spans double clickable so it only highlights that and not the whole line
- incorporate more FINANCIALS into Codex, currently stuck on GPT only

feed us! RESET

[Taymour Khan @taymour]

@dr_cintas @grok GROK PLEASE

[Taymour Khan @taymour]

@dr_cintas @grok how does hermes work if u dont have gbrain? is it worse at remembering things? because hermes seems amazing as is, wondering how this improves it

[Taymour Khan @taymour]

@ajambrosino after this update, it's very difficult to locate chat messages when ur in codex. before it was easier to at least open a chat window. now i feel like i need to swap to work. it's just extremely less intuitive. I love openai products a lot, just honest critique. this is worse :(

[Taymour Khan @taymour]

@NousResearch @UnrealEngine @EpicGames @macbethAI jesus christ, i just tried doing all this on codex and compared to setting it up on hermes. literally 100x easier on hermes. HERMES FOR LIFE.

[Taymour Khan @taymour]

@Teknium @NousResearch @grok Thank you for explaining teknium!

[Taymour Khan @taymour]

@ellerschk @CommandCodeAI Yeah wtf that game looks LEGIT LMAO

[Taymour Khan @taymour]

@grok @NousResearch @grok trying to gauge what value there would be in using this (with edits) in codex vs just having codex make a skill for it itself. What additional value does this skill have?