Appreciate all our investors @DraperVC, Amand Ventures and our angel investors who have supported us from the time we only had an idea. Draper Associates has invested about ~1.114million USD since 2020. More about our journey with Tim who tested our technology every year before investing more and other public and social partners in this thread: x.com/ye_dennis/status/19565… (Only investments 50k USD or more are listed on our @crunchbase: crunchbase.com/organization/…)

That's a wrap. 8 hours of Virtuals goodness. Thanks to all of the ~40 teams that presented, the hundreds of you who showed up to listen, and my brother @goon_crypto! It's clear we have something very special here within the Virtuals ecosystem. It was a pleasure to showcase the best of it today for our 50th episode celebration.

Virtuals Weekly Episode 50 Recap! ➥ 38 teams, over 1600 listeners, across just over 8 hours. We smashed all records with Episode 50 becoming the biggest show the @virtuals_io ecosystem has ever seen. ➥ Thank you to all of the teams that presented. We had some amazing updates and alpha from the very best that Virtuals has to offer. It was a true celebration of what is great about the ecosystem. ➥ To the audience who came along and supported, we appreciate you. You make this show what it is. It speaks volumes about the health of the ecosystem that so many people tuned in. ➥ Thanks also to @DonJohnsonSays who was there to open and close the show. We appreciate the support we get from Don and the Virtuals team. It pushes us to continue putting out this content for the ecosystem and providing the platform for teams. ➥ The milestone is the beginning of a new phase for Virtuals Weekly. We've got big plans and are looking forward to the next 50 episodes. Rumours of a 24 hour stream for episode 100 next year! ↛. As always, you can listen back to the recording in the comments. We've also got a stacked show booked for epsiode 51 next week at the earlier time of Monday 10pm UTC. Set your reminders in the comments too.

Most people talk about model quality. They forget the data that trains it. @Repponetwork is about fixing the data side of AI. ↛. The problem is simple. AI needs high-quality human preference data. That data is controlled by a few large labs. ➠ Hard to access ➠ Expensive ➠ No transparency The world needs an open way to produce this data. ↛. Reppo turns AI training data into a market. Publishers create AI content. Voters stake and vote on what they believe is best. ➠ Publishers post outputs (text, images, video, agents) ➠ Voters lock $REPPO ➥ get $veREPPO ➥ vote ➠ Rewards flow based on collective preference Quality rises because the crowd prices it. ↛. Voting is not clout. It is a prediction market. You are rewarded when your judgment proves right. ➠ 50% of weekly emissions go to creators ➠ 50% go to voters who backed them ➠ Commit-reveal hides results until the epoch closes Clear incentives. No guessing games. ↛. The output is not hype. It is data. Real training signals for RLHF (Reinforcement Learning from Human Feedback) and DPO (Direct Preference Optimization) ➠ Ranking ➠ Pairwise comparison ➠ Edits and corrections ➠ Expert feedback This data helps models behave closer to human expectations.  ↛. Subnets take it further. Anyone can spin up a data environment. ➠ Public: open contribution ➠ Private: controlled for companies or research teams ➠ Each subnet collects domain-specific data This scales across fields like law, medicine, design, robotics.  ↛. Solver Nodes come next. They help route and monetize the data. ➠ Nodes source data ➠ Sell access to AI teams ➠ Pay fees back to the network This ties usage to value. ↛. In short: Reppo is not a model. It is the coordination layer for training data. ➠ Publishers earn from quality ➠ Voters earn from judgment ➠ AI builders gain aligned data ➠ The network grows from participation ↛. The bet: The world will need far more aligned preference data than any lab can source alone. If you believe in open AI, you watch Reppo closely. Launching on @virtuals_io Monday 17th November!