Co-Founder, Polyshares.com // Enterprise-Focused Data Lab // Exited $35M annual rev Solar Company (2025)
- Tweets 1,783
- Following 1,569
- Followers 1,272
- Likes 6,222
Pinned Tweet
If you want to get a piece of this, we just helped a roofing company get a $450k offer for their De-Identified Data.
These labs need your workflows to train their models and are paying high dollar right now ($100k-$1M)
Dm me if you want to get a quote on your data.
Wow, seems like Google is buying Spirit Airlines' enterprise data for $10m (outbidding Mercor at $7.5m).
Basically includes every internal document, email, workflow, and codebase for a once $6B company.
Honestly, $10m for 34 years of operational data really seems like a steal.
23
11
420
128,789
I used Jev as a verb in casual conversation today
Do I need to pack my bags and just move to SF at this point
2
222
Hey @chesscom can we work on adding a feature where I can pick out a point in my game where I found a great move and create it into a puzzle to share with friends.
Either that or a "Highlight Reel" style gif I can share
13
390
Don’t get just one offer.
$600k is a lot of money, but the just like the old saying goes “Your Price, My terms”
Firms like Micro1, will give you exploding offers that expire before they’ve even seen the data themselves.
Their offer will include full exclusivity for perpetuity.
I can think back through history at points in the Oil/Mineral Rights or a Record Label where taking a perpetual exclusivity for an upfront check was a very expensive mistake for a lot of sellers.
If you want a competing offer that doesn’t blow up in your face, DM Me.
We were offered like $600k to sell our company data to Micro1. 100% not doing it.
But it does interest me like crazy.
What a crazy business model and idea.
If you don't know, here's all I know about them: they go to private companies. They ask for your slack, notion, email info. Offer you 6-7 figures. Somehow anonymize that data and sell it to Openai, etc.
Seems like Micro1 has grown to 9 figures in revenue very very fast.
What else am I missing here. I get the value of this. But doesn't seem right to me.
4
47
42,731
Polyshares will not sell data to Chinese Labs or anyone found to be sub-licensing to Chinese Labs
It's a well-known secret in the valley that Chinese AI labs are buying training data and "compute credits" from startups
Training data is an extremely hot commodity these days, companies are paying hundreds of thousands (sometimes $2M+) to buy Slack messages + google drive + other docs from startups
Many of these training brokers are pretty open about selling right to Chinese labs
This happens even more explicitly through excess compute credits. YC startups get $500k of tokens from OpenAI & Anthropic and Chinese labs are sending s3 buckets of prompts to "buy out" the token credits from morally ambiguous founders
nitter.net/dennishegstad/status/2…
nitter.net/armon_sho/status/20832…
nitter.net/annatonger/status/2085…
12
1,297
Model H was being rewarded with Fruit whenever she ate her vegetables.
Model H has now learned to throw her vegetables on the floor to farm the fruit reward.
Has anybody ran into reward hacking when training their little model?
10
434
If we see a slow down on Frontier research, would this then reposition the most technical and skilled researchers towards commercial application model training.
1
3
458
Seems like a great Research Finding for companies positioned to provide premium off-market data sources to researchers.
Great Data 🤝 Great Researchers
@Deepseek_ai demonstrated improving data quality delivers far greater gains than developing new algorithms.
Good data is the bottleneck.
1
265
Ryan Lock (🤠/🤖) retweeted
This is notable. DeepSeek, a lab usually first to pioneer novel algorithms and architectures, is saying that at this point, the ROI of improving data quality far exceeds that of working on novel post-training algorithms.
I think this has already been true for some time for non-lab practitioners. If you're doing llm post-training, 80% of your effort should go into looking at your data.
This means:
- Hiring experts to dig through your RL tasks
- Sifting through rollouts and sft data by hand to remove suspicious samples. Make sure all tasks are actually passable.
- Making sure your data is diverse in both difficulty and category.
🚀 Introducing DeepSeek-V4.1-Flash: smarter, faster, more efficient.
🔹 Introducing the smallest model in our new architecture family, with native visual understanding.
🔹 Designed for greater capability, faster inference, higher throughput, and scaling to larger models.
1/6
57
151
1,733
209,325
On a scale from 1-10 (1 being Anthropic, 10 being Elon) where do you see the world in 10 years?
2
237
Polyshares is having extraordinary success with sourcing specific data sets.
A recent lab partner needed a data set from an industry with a specific software stack and within 48 hours we had an owner ready to license.
Polyshares is becoming a competitive advantage for labs to find the data they need and get it before their competitors do.
Excited to keep growing with our partners and delivering results.
3
17
5,447
This has been a great few weeks for data. Since the Google acquisition of Spirit Airlines, we have booked over 100+ calls with business owners looking to get in on licensing their data for AI Research.
There is only going to be a moment of time that your De-Identified data is as valuable as it is today.
polyshares.com/intake
Find out what yours is worth today.
1
5
803
This is why I’m bullish that every founder should be using some of their fundraising money to buy ownership into a high-end restaurant.
Instant high signal to potential clients/investors to invite them to “Your” restaurant.
Pro Tip: Make sure members of the staff come over at least once during the dinner to thank you for some obscure favor that you did for them, but can’t discuss in detail.
A GTM pattern I’m seeing among some of the best young founders:
They’re using highly curated dinners to build customer communities.
One founder selling into health systems is hosting Michelin dinners for health-system C-suites.
Another is convening fintech leaders in New York.
It’s targeted, high-trust, and working extremely well.
1
7
669
This is going to be the new GTM strategy. Launch a 404 game with your website.
made a little game for myself and the gameplay is too good to keep, so now it’s the new 404 for @iskra_graphics
you can play it right now at iskra.graphics/404
315
AI Researchers being reported on like NBA Players is my new favorite thing
this just in: Barret Zoph, who dramatically left Thinking Machines Lab to go to OpenAI, is now going to Google as a VP of research on RL/posttraining
completing the boomerang of Google --> OpenAI --> TML --> OpenAI --> Google
scoop with @MeghanBobrowsky wsj.com/tech/ai/thinking-mac…
1
2
385
