3+ years of building our thought -> machine interface (BCI).
we are approaching a GPT-3 moment for AI Brain models.
lots more to come as we roll out updates.
for now, check out our new site: https://t.co/nRZU2KjfFI
i spotted @rahulchhabra07 and @abanerjee99 building @sabi in early may 2024. i wanted to invest. problem was sabi wasn’t incorporated and didn’t have a bank account yet. so i planned to meet rahul as soon as they did. on may 13, 2024, just 3 hours after sabi’s incorporation was inked and banking was set up, i met with rahul. i had him share the docs and wire info and within 20 minutes of our meeting, @maniacvc had signed and wired sabi its first check.
but the sabi story starts earlier for me. it was december 2023, NeurIPS conference. a paper i had read on arXiv in back in may 2023, “MindEye” was being spotlighted. MindEye showed that fMRI brain activity could be mapped to any high dimensional latent space, which meant enabling image reconstruction using generative models that accept embedding from this latent space. MindEye retrieved exact original image even among similar candidates showing it was possible for brain embeddings to retain fine-grained image specific information. in other words? brain ai interface was possible. BCI was here. something i had first gotten excited about as a kid, fueled even more by my work in deep learning and neural nets at the university of toronto in 2014. MindEye was poster spot 416, the area code of toronto and birthplace of deep learning, i took that as a sign.
since the conference i kept tracking the researchers behind MindEye (well, my agents did) and when i saw that atmadeep, MindEye’s second author, was working with rahul to commercialize the research but now focused on EEG signals over fMRI, i reached out immediately. and now we’re back to where this post started, may 2024. wiring the first check into sabi. and since that day maniac ventures has backed them again and again with a 2nd and 3rd investment.
@vkhosla put it well, “very simply, thought-to-computer input is a sexy idea.” but i think it’s beyond sexy. and so does vinod i’m sure. i bet that brain ai interface will be the primary way we interact with computers. and the results are already in. sabi can already transcribe people’s thoughts with up to 77% accuracy (!!!!) for those who have been using it for more than two weeks. read that again. BCI is here. and talk about perfect timing. using sabi you can think what you want done, and it happens. thought to software, thought to hardware.
computers are turning into agents. and thought is the native way to talk to them. sabi will be presenting at CES, and i think the world will catch on more to what they are up to. as for me, i’m just happy i got to meet rahul and atmadeep right at the start of their journey. i have a feeling the world will be talking a lot about sabi in the months, years and decades to come. as rahul puts it, “models are getting better and better and there’s a fear that they’ll not be aligned with humans. if AI can understand human logic and human feelings, it’s a lot more likely to do what we want.”
i had a feeling the day i wired sabi its first check that it would be a day i’d never forget, and i’m a dad who takes lots of photos so i was sure to take a picture of the exact moment. it’s the one below. the mural behind us had the words, “don’t forget your keys” scrawled on it. very appropriate. with sabi, chances are good you won’t.
rahul told forbes that “a computer is no longer a set of keyboard, mouse and monitor. they are much more about AI models that are doing things for you, and they could just be with you as you’re walking.” amen. just think.
This company is building mind reading technology for defense applications...
Read the full company breakdown here: https://t.co/CVPVf99pfO
@TelepathTech, founded by @NoahShamus and @ElleryBuntel, backed by @joinef and Transpose Platform.
“MGM put a 5$ limit on me because of arbitrage betting”
That’s what @NoahShamus told me the first time I met him. He was working on low-latency solutions at a fintech. Very quickly we knew that such an obsession about a topic and willingness to do whatever it took to beat the bookmakers could be applied to other problems and industries. A few weeks later, Noah joined the cohort and met @ElleryBuntel
Ellery was obsessed with brain-computer interfaces and had been doing research on the topic for the past three years as part of his phd. I met him for the first time in San Francisco at the augmentation lab demo day where he was sharing some of his work on image reconstruction using EEG data. Ellery dropped out of his PhD, and joined EF.
After meeting in our first US cohort in New York they built Telepathic Technologies, a noninvasive brain-computer interface to decode language-related thought in the brain. Check out their pitch below 🚀
The next cohort kicks off June 17th in SF. Places are almost filled, but with the influx of messages and interest we received post demo day, we’ve decided to open applications for one week only. If like Noah and Ellery you want to build with the best founders, in the most intense and founder-friendly environment, apply before May 10th (link in comments)
6/ @TelepathTech are building non-invasive neural links to empower the mind of the modern warfighter. Cofounded by @NoahShamus and @ElleryBuntel just a few months ago, @TelepathTech can now identify an individual word in the brain with 75% accuracy. They've tripled the accuracy in just 3 months by using non-invasive neural links.
We invited 150+ Neuro builders to showcase what’s possible with BCI tech
This was the largest Neuro Hackathon I’ve seen, organized with @neurosity and @TelepathTech
Live product demos. No bullshit.
Here are the best projects:
Thrilled to be in SF building some awesome neurotech with @NoahShamus for our company @TelepathTech!
Feel free to reach out if you want to chat about innovative neurotech.