Eghosa Omoigui

'Sometimes it is the people no one imagines anything of who do the things that no one can imagine' - Alan #Turing, The Imitation Game

@MarcACaputo What a beautiful tribute to a remarkable adventurer and storyteller. Condolences. May he rest in peace. archive.ph/Lysg0

Philip Caputo wrote a best-selling classic about Vietnam, shared a Pulitzer for uncovering Chicago’s voting fraud, was captured by militants in Lebanon & shot by others, covered wars in Afghanistan to Africa, hunted big game & caught bigger fish RIP dad facebook.com/share/p/1EwYov…

@WillManidis to restore dignity, in mankind and of its labor.

@WillManidis Reading this, I'm reminded of Marshall Ney's remarkable rearguard action in Napoleon's 1812 Moscow retreat. We learned catapulting to an end game req's rearguard strategy. “Soldiers, when I give the command to fire, fire straight at my heart. Wait for the order." - Marshall Ney

Rewiring #AI in 2026: The Convergence of Continual Learning, Decentralization, and Collective Intelligence substratesm.com/post/2026-02-0…

Arguably nonconsensus.

@jaltma @gradypb @Alfred_Lin This was excellent.

I sat down with Sequoia's new stewards, @gradypb and @Alfred_Lin. Maybe a bit selfishly, I spent almost the entire time asking them about the tactics of how to do venture capital well. I got a ton out of it and hope you do too. Search Uncapped on youtube, spotify, or apple to listen there. (0:00) Intro (1:01) Mindset as stewards (4:30) The business of outliers (6:27) Managing inputs in venture (12:11) Sourcing coverage goals (17:57) Seeing the right companies (22:36) Proprietary map of talent (24:39) What a single engineer can do (29:06) Picking winners with conviction (36:26) Coaching asymmetry into picking (43:16) Disagreeing with younger investors (46:45) Frameworks on picking (53:20) What it takes to win (58:32) How to start working with a founder (1:02:59) Proudest board roles (1:06:12) Looking ahead

@_clarktang The skepticism re ODI (on-device inference) reminds me of when AAPL released the MBA. No one believed a laptop could be slimmed down that much. INTC’s innovation took a very different approach. My point is that spending time thinking it’s impossible makes it so. Not ODI but MDI.

Everyone always asks about edge inference - so just gonna leave this here… to build on Zephyr’s points Was an investment theme we were quite hopeful on - but when we looked across Silicon Valley, Taiwan, Korea - all checks were reasonably negative Question before used to be model size / quality but quantization and distillation have helped that a lot. Quite a decent model you can fit in 3-7b parameter envelope now BIGGEST problem to all of this adoption that no one talks about is battery life. In DC you have constant power, on device you take a 8 hour mixed use battery and decrease that to 1-2 hours for an on device LLMs to spit out 5 tokens / s. Can’t fit a larger battery on the form factor we are familiar with, so will have to strap a battery pack lol Btw each $100 increase in BOM will be an increase of ASP of $250 in premium device. I don’t know anyone who would accept that tradeoff so they can run Llama 7b on device Finally if you really want fast inference of 7b models you might as well use Cerebras or Groq as an cloud call - will be much faster than waiting for your on device phone to spit those answers out Overall remain bullish and optimistic for new interface of computing. But the form factor will have to look different, the use cases will have to be discovered / invented

@uninsightful @Alfred_Lin led the $127m Series C round in Mar 2016. Rumor has it he invested ~$100m @ ~$600m pre. Stepped up when the VC mkt stopped believing in the DASH team & opportunity. For those @ home, $10b+ in gross receipts assuming hold till ATH. So yes, my fave VC is doing it again.

more and more early-stage venture capital firms are stepping into the public markets. I strongly believe that this is a trend that is here to stay the nimble large platforms have been doing this for a long time (Ribbit, Greenoaks, Thrive, etc) and many other firms of different sizes have joined them related: on a dollar basis, the majority of capital deployed by crypto "early-stage VC" firms in the last 3 years has gone into liquids most notably, in most of these cases, it is the same investment principals that are touching both early-stage privates and publics (even for myself, I spend ~10% of my time @usv on publics/liquids) at first glance, this looks like scope creep, but there are actually several good reasons for this. @pchopra28 does an excellent job articulating some of them in his attached post:

@Espngreeny RIP Micheal 'Sugar' Ray Richardson. This interview will remain immortalized.

This makes me so sad. Michael Ray was Showtime before that was a thing, he was the among the most flashy, gifted, and flat out watchable players in the NBA. RIP to a different kind of legend; those of us who remember will never forget.

@raizamrtn Pincer strike and smartphone in the middle.

If I was Apple, this is what I would go all in on- Audio as the natural interface for AI, especially with airpods as the hardware form factor for two reasons: 1) The amount of people with airpods in all day (already) makes it the easiest surface to launch to. You don't need to convince the user to buy new hardware - just find a way to make your product work seamlessly, in new and delightful ways. It's hard enough to convince people to try new things - so it's uphill to assume hardware + software behavioral changes at once. There's a ton of stuff that can be built for audio that are totally new product paradigms, and IMO not enough people are thinking about it. I could list 10-15 more things I'd be exploring in this space that we're barely scratching the surface today. If you're interested in building hardware, start here anyway because you'll learn about exactly what it is people want along the way and will improve your hypotheses. 2) People will balk at glasses but they won't at airpods. It's not weird to be talking to yourself when you've got your airpods in and people won't care if you're talking to your mom or an AI. Literally the same experientially and already (mostly) socially acceptable. This makes it an extremely interesting learning surface for new interactions with AI because the user isn't screen-first: just a totally different mindset for the user. It represents a totally different cohort of queries and actions than what we can learn via other modalities. So even the data play is valuable.

@raizamrtn I'd argue in addition, and thus a promise fulfilled, that it's a two-sided zero-cognitive-load experience. B side: How do I know what I don't know to ask? x.com/EghosaO/status…

I really believe that the next turn of AI is personalization and proactivity, with one massive caveat: These two things are only powerful if they are "subtractive." Let me explain. Most “personalized” products today add more: more recommendations, more alerts, more surface area. True personalization should do the opposite. It should quietly take things away. The same goes for proactivity. A proactive system shouldn’t interrupt you, it should preempt friction. It should remove steps, not insert itself. Since we launched Huxe a month ago, I've been talking daily to our users. Unanimously, each person describes the best quality of the app as being subtractive: it takes away something as its core value. "I listen to it first thing in the morning instead of opening my mail app" "Instead of opening a bunch of tabs, I just listen to the news" "I always look up this niche topic to keep track of it, but now I just created a station about it" In these quotes I'm hearing what I've always believed about good products, too: there’s a real usefulness in things that unburden us, especially from tasks we’ve always believed we had to do, but never actually liked doing. So as we build the app more, we're really embracing the art of subtraction. People want less: fewer tabs, fewer decisions, less friction. Not just an app that does everything for you, rather an app that removes everything that gets in your way.

nowacentre.com/in-loving-memo… Tough year. Grateful to all for the support. Pls extend support (earmarked for GBM) to the NBTS / Brain Tumor Investment Fund braintumorinvestmentfund.org/portfolio/, Yemaachi Biotech yemaachi.com & Glioblastoma Research Organization gbmresearch.org

Nowa Omoigui, MD (1959-2021) and Moira Omoigui (1962-2025). Sharing memories of happy times. May their souls RIP.

@jsngr Manus does the whole async pulse thing really well.

this is one of those obvious in hindsight, everyone will copy it kinda things one of the best properties of agents is that they can work async in the background. live activities is a great way to keep tabs on work happening while you watch

@iramneek Unroll @threadreaderapp

1/17 Ed Thorp: You know, Nassim, if I had a dollar for every hedge fund that blew up pretending variance is risk, I’d have Bridgewater’s AUM. And that’s before the performance fee.

@iramneek Unroll @threadreaderapp

17/17 Thorp: (raising his teacup) Cheers to that. Survival isn’t sexy, but compounding is the only revenge worth having.

11/17 Thorp: And when they do jump, the ones who sized wrong don’t just lose - they vanish. There’s no ‘mean reversion’ when you’re out of the game. That’s what Kelly taught me, and Pascal before him - there’s no recovery from infinity loss.

Enjoyed this article on Max Levchin & Affirm. semafor.com/article/03/20/…

Told folks on WhatsApp months ago that the GBH wounding that Deepseek was gonna deliver by end of Q1 was doing press up in the gym while US/EU teams were shouting about the initial open source release. The bigly funded LLMs should press charges, tbh. This is full on assault.

@asemota lol. wait till the US offers sanctions relief et al. this process is very sumhau.

We now know who doesn't want the war to end.

@jadeyversat Sweet n' sour, huh? Clever. Very clever. I posed a related question to @eajene. x.com/EghosaO/status…

@EghosaO This is likely an own goal to China if they fail to rechannel the energy of those low skilled factory workers that'll be rendered redundant upon full automation with AI enablement. It's a sweet -sour situation ☺️

Automation shock. Africa where art thou? “China exploited its comparative advantage over recent decades in having an abundant labour force & became the dominant manufacturer globally of labour intensive goods. That game is now up.” - Frederic Neumann, chief Asia economist, HSBC.