Delighted to share that DSV took an OTC position in @adtao_ppcrebel (SN21) ahead of Proof of Talk.
We got in early because we know where this business is heading. Combining the power of Bittensor with the size of the Google Ads market.
AdTAO is real advertising intelligence built on Bittensor, led by an operator with 13 years in the game who has built and sold multiple ad businesses.
That’s why we wanted to lock in exposure for DSV investors.
Today the team announced their first buyback. They've already got fiat revenue and they are using some of it to back their own subnet alpha.
Proud to back AdTAO.
Hey everyone -we're a day late here as I had some tech issue yesterday but I can confirm that we after less than 3 weeks on mainnet we have completed our first buyback. $1k, 5TAO approx ha been bought and burned. https://t.co/A7zqDNkRzX
SN21 EPOCH W21 Initial Recap:
Yesterday saw our first epoch on mainnet complete. Some things worked well. Others not so well. Here’s our initial assessment:
Worked well
- 16 miners submitted scores successfully
- 3 miners achieved an elite status result - that’s a real tangible deliverable
- Everything verifiable on chain
- Leaderboard live https://t.co/XdtNTZ8cSU so status can be verified easily
Issues
- Leaderboard had to be significantly updated on validation close to be more useful. This slowed down miners from getting results if they were unfamiliar with checking on chain
- 10 miners did not register (instructions in readme) and did not receive a score
- Some validators did not switch weights correctly so even though we set a 75% burn rate, miner emissions were much lower than expected - this is despite us working with validators to get them to weight properly (note - we cannot control validators).
Actions
- We’re improving our submission error messaging today so that if you haven’t registered but try to submit, you get a clear error message so you know to register and resubmit
- We’re continuing to push validators to correct their weights. We have Tensora supporting us in this action. We’ll keep pushing - but we don’t have control
- We’re going to run another BID_STRATEGY epoch next week W23, so that all miners who did work that wasn’t scored can resubmit. Providing validators make the required changes everyone will receive their proper, ranked reward for W23. It will use a new datasest for fairness but robust solutions can be resubmitted unaltered. Clearly, we can’t guarantee ranking but this seems a fair way to allow everyone to get their existing work scored and rewarded.
@TAOTemplar@TaoIsTheKey Great shouts and thanks for all the feedback. We’ll get it all done. I’m currently on vacation several thousand miles from my office so trying to figure out if I can do it from here or whether it has to wait till I get back
@AnthonyHigman Increasingly shifting to voice while walking. Typed at desk. I’ll bet the correlation with voice and seemingly insane long queries is very high.
So this week was our first week live on mainnet so now that hurdle is handled we can ship more product updates.
At the risk of pre announcing something we will be releasing our first big product a. Oh cement before the month is over.
Can’t wait!!
And in the interim if anyone takes a look at our repo you’ll see it’s far from quiet in there https://t.co/qxcrqO1gQb
@NeptuneMoon It’s become an exclusion game now. You “steer” with a keyword or search theme or however they rebrand and then you heavily exclude everything you don’t want in the direction you go.
Include=small
Exclude = big
@TAOTemplar Great. That’s why I asked. I need to figure out how to correct the data. We’ve committed a lot of code in the last couple of weeks so need to get the data fixed.
It’s been a tonne of work but exciting to get Adtao live and generating predictions. Miners - can’t wait to see what you do.
We can build something fantastic here.
SN21 incentive mechanism is live on Bittensor mainnet.
AdTAO is a Google Ads prediction subnet. Miners predict counterfactual outcomes of advertising interventions across thousands of ad accounts.
Phase 1 is Search campaign-level episodes, the cleanest signal in the pool.
Reward structure is simple by design:
→ Beat predict-zero or earn nothing
→ Elite tier (top 20%): 60% of emissions
→ Competitive (next 40%): 30%
→ Participating (bottom 40%): 10%
Registration is open. This is a prediction model problem; Google Ads expertise is an advantage, not a requirement.
Docs: https://t.co/JcjjYklrGT