Knoww.ai retweeted
WHAT AM I EVEN DOING?( Building category creator model)
The LLMs are trained for extracting intelligence from unstructured data.
TFM is built to extract intelligence from structured data.
I am building second.
Businesses are dependent on structured data. I extract intelligence in seconds and mins where the existing tools like XGBOOST takes hours.
We help DataScience teams to become faster the way LLMs helped frontend and backend folks became superfast!
Go to knoww.ai and fill the form with work email and let us help you with extracting intelligence from your own data.
Knoww.ai retweeted
As a global team of 7 people from SF 🇺🇸 building Tabular Foundation Model, till now we’ve:
- Built a completely fresh community of 10K+ people
- Talked to more than 500 people on chat in DMs who showed interest
- Had gmeet with at least 50 people who were genuinely had done the work
- Closed 4 positions on intern level who are kick-ass, absolute cracked and love what they do
- Got almost 2.5M impression in the last 40 days around our foundation model + product
- Closed few more discussions around the foundation model’s prospective customers to use our product
NOTE: If you want to extract future predictions around your business’ most pressing problems e.g. churn, revenue, leakage, preemptive maintenance and etc, do reach out. We’re onboarding customers. We won’t charge you until you get clear RoI. Promise! But you’ll have to start with signing of the LoIs with us.
Reachout by filling the form on : knoww.ai
Knoww.ai retweeted
You have asked me so many times that what do I read for the mechanistic interpretability and blogs. So here you go. You can also get on top of the mechanistic interpretability discussion just by reading these three sources:
- Neural Nets are already clean by Nick Cammarata already-clean.goodfire.com/
- Neurons as an interpretablebasis : goodfire.com/research/neuron…
- Neuron Communities : goodfire.com/research/neuron…
Also - We’re onboarding customers on our tabular foundation model :
1. Go to the site knoww.ai
2. Fill the form.
3. Get the access to the model
4. Start testing your hypothesis
Knoww.ai retweeted
As a global team of 7 people from SF 🇺🇸 building Tabular Foundation Model, till now we’ve:
• Got 3 acquisition offers: One Listed and 2 Private Companies.
• 2 co-founder offers.
• Ran around 15+ experiments and collecting different observations.
• Became 6th globally on internal benchmarks of running DevTasks of TabArena.
• Matched the research of the Stanford, NUS & other pioneering companies innovating in this domain.
• Innovating on top of the work of the giants.
• Got inbound of the top of the talent from Stanford, Berkeley, IIT-M, IIT-B, ISRO, DRDO, GROW.
• Got 10+ customer conversation of which 7 are currently testing the product.
Knoww.ai retweeted
youtu.be/hAT72lwY5uw?si=WzkY…
I recorded my first research paper explanation video. The goal is to enable more people to get in this field and remove this fear mongering from the world that its so hard to understand and they can’t get in this field.
I want you all to pull up your socks and tag along with me now in this journey! I am gonna make sure that you learn what I learn! Don’t fall behind!
Just roll up your sleeves and get to the job!
You can watch it on 1.5X.
Knoww.ai retweeted
TFMs- Tabular foundation Models needs to be re-looked at if we want the data to give us result faster.
Be on the lookout for the innovation that we will drop. Or call it progression of our efforts.
Knoww.ai retweeted
What if you can fetch a single mathematical equation from the whole Neural Network which is also giving you same result that the Neural Net has given you? Can the outcome become explainable?
and Won't that equation become equally potent as the whole Neural Net?
After a month of efforts, I'm happy to share that we were able to reproduce an equation which can replace the whole Neural Net and still fare well in the comparison.
Her's a brief demo that I recorded to show the comparison of the :
1. equation
2. with the Linear Regression that we're replacing with
3. the TFMs
We're contributing in understanding the neural network's outcome from explainability perspective. Its like interpreting the model's brain. Here's the video:
We've onboarded another intern. He showed his mettle by picking up the tabular domain at blazing speed. He built a v0.1 of our model and ran it against the benchmarks.
We're 6th by Elo on our own reruns of TabArena Lite (DEV tasks only, unofficial, not on the official leaderboard yet). And we're hillclimbing to the Pareto frontier from here.
Meet Tabular Multi Latent Knowledge Space with Longitudinal Intelligence and Accurate Recall Model- i.e. TAB_MLKS_LIAR_M_1:
First of many experiment outcomes. Thank you for your attention to this matter.
Knoww.ai retweeted
We just hired an intern who have worked on diffusion models in CERN. The goal is to have super dense acumen folks who are great at pattern matching and thinking from first principles.
First win! Stay and tag along. You are having a front row seat in building a generational company on Tabular FM.
Knoww.ai retweeted
Hello 👋,
Yes its a case of “Tabular foundation model” and I have started hill-climbing it already. The goal is to be SOTA in these fraud/underwriting use cases.
xitter.gq/heyvixon/status/210920…
Knoww.ai retweeted
.@rauchg ships Fable 5.1 on Vercel AI Gateway.
and @Knoww_App immediately asks the next question: when does Fable ship again?
Market says:
61% by Dec 31
33% by Oct 31
This is the internet I want.
Just Knoww it.
Knoww.ai retweeted
This is why I love building @Knoww_App.
@smtgpt posts a fascinating India vs US wealth comparison.
Knoww.app immediately turns it into the next question:
Bitcoin, Gold or S&P 500… what wins 2026?
You read the take.
Then you see the odds.
Feels way more alive than just scrolling opinions.
More Americans hold Bitcoin than Gold. In India, the case is the exact opposite: ~75% of adults (~700 million people) hold gold, vs ~10% (~97 million) who hold crypto.
Having grown up in India and spent 8+ years building in fintech, this makes sense to me.
In the US, ~62% of adults have investing accounts and Crypto is available on most of the fintech apps people use to buy stocks. In India, for a long time, people invested in physical assets like gold and real estate first and then looked at other avenues.
That's changing though. A generation raised on UPI already treats digital as default for paying, and increasingly for investing and trading with VDA gaining strong consideration as an asset class. Gold took generations to reach 700 million Indians. Crypto went from zero to roughly 100 million users in about ten years.
It'll be interesting to see how the data below changes in the next 10 years.
Knoww.ai retweeted
okay this is actually such a sick @Knoww_App demo.
@notthreadguy posted a take on capital gains.
Knoww understood the topic and pulled the relevant live market underneath it automatically.
this is exactly how prediction markets should feel.
Knoww.ai retweeted
“Strait of Trump” is exactly the kind of headline where I need odds immediately 😭
@Knoww_App pulled up the market right under the post:
Yes 7% / No 93%
Way better than opening another tab just to figure out if people actually believe it.
@PolymarketTrade @Polymarket @shayne_coplan
Knoww.ai retweeted
.@ankurnagpal posts about the insane wealth creation from an Anthropic IPO.
My next question: when does it actually IPO?
@Knoww_App: 95% by Dec 2026.
One tweet → one question → market already has a price for it.
Knoww.ai retweeted
.@hosseeb says Anthropic IPOs first.
Prediction markets say 95% Anthropic. 5% OpenAI.
This is the internet I want.
Someone posts a prediction → Knoww instantly shows you what people are actually betting on.
@Polymarket @shayne_coplan @devjoshstevens @PolymarketTrade
The Anthropic & OpenAI IPOs will be the largest ever liquidity transfer from global institutions to SF futurists.
And what will they plow their money into? I expect the two most concentrated beneficiaries will be SF real estate, and crypto.
Employee lockups expire usually 180 days out from IPO. Anthropic IPOing by mid-October, so expect this to start in April 2027 and continue with OAI's IPO next year.
Watch this space.
Knoww.ai retweeted
Just Knoww it!
Bro dropped a GPT-5.6 Sol take and @Knoww_App casually pulled up what the market is betting on next. 🔥
Every opinion on the internet probably has a market behind it. We’re just bringing that market to you.
Just Knoww it.
@Polymarket @shayne_coplan @devjoshstevens
Knoww.ai retweeted
Macro Twitter gives you the thesis.
@Knoww_App gives you the odds.
@akshaybd talks GDP growth → the live GDP market appears right underneath.
Just Knoww it. ⚡️
Knoww.ai retweeted
This is my favourite part of what we’re building with @Knoww_App.
@balajis is talking about Kazakhstan, investment and Solana.
Knoww understands the context and brings the relevant Solana market to the post.
Don’t search for markets. Let markets find you.
Knoww.ai retweeted
Replying to @hosseeb
Market seems pretty convinced you’re right.
Anthropic 95% vs OpenAI 5% to IPO first.
This is what we’re building with @Knoww_App. When someone makes a prediction on X, bring the actual market odds right underneath it.
Knoww.ai retweeted
Crypto Twitter has enough takes.
I want the market underneath the take.
@hosseeb mentions Hyperliquid → @Knoww_App instantly surfaces what people are actually betting on.
That’s the internet we’re building.
































