𝔻𝕣 𝕎𝕚𝕫

Lagos just joined the mesh. Pitched Pollinet at a co-working space today, and watched real people download the Pollistem app, set up their nodes, and become part of the network. These aren't users. They're infrastructure. Every phone in that room is now relaying transactions for Solana wallets and dapps. The network just got stronger. Join the Network and you the referral code: V59HYE play.google.com/store/apps/det… @SuperteamNG

just found this hidden gem on @base if you track AI narratives you need to see how @gitlawb replaces traditional backends with decentralized primitives key highlights ■ anyone can join the network without passwords because authentication relies entirely on HTTP signatures ■ the current version already hosts thousands of repositories with steady push volume and zero paid marketing ■ developers post tasks in native tokens and agents automatically submit PRs to claim rewards ■ it offers insane demo ability since you can watch an agent fork a repo and write code live they building an ultimate decentralized hub where AI agents and developers ship actual products together seamlessly.

I invested in @Minima_Global because the thesis is architectural, not speculative. The EU AI Act becomes enforceable for high-risk systems on August 2, 2026. Every autonomous machine operating in Europe (drones, industrial robots, medical devices, embedded AI) will need tamper-proof audit trails that regulators can independently verify. The compliance infrastructure the market has built so far assumes a cloud server and a persistent internet connection. Most of the machines that need it most have neither. Minima runs a full blockchain node in 300 megabytes, on a microprocessor, on the device itself. No gas fees. Quantum-resistant cryptography. The partnerships are not slideware. @Arm contributed security-rich IP to build the first blockchain-on-chip. @Siemens is distributing through automotive, robotics, energy and healthcare channels. @Fetch_ai has integrated the compliance layer into Agentverse for autonomous agents. @unisouthampton validated the architecture in live drone flight with a 500x performance gain. The Minima market cap sits around 3.7 million dollars. A single enterprise AI compliance implementation costs more than that. I have looked at this from the regulatory side, the technical architecture, the partnership validation, and the market pricing. The asymmetry between what has been built and what has been priced in is the largest I have found in the infrastructure layer. This is not financial advice, but I am a buyer. Minima is listed on @MEXC and @BitMartExchange and @XTexchange

could we predict the second world war if we had LLMs back then? i gave 34 of my agents running on @miroshark_ simulation engine a new model called talkie, trained only w/ pre-1930 data. asking the question 'Does Germany's 1930 Reichstag see N*zi Party seats exceed 100 by year-end, accelerating democratic erosion?', my agents Mahatma Gandhi, Joseph Goebbels & Himoto (?) started to post on X, Reddit & bet on Polymarket they created three markets: - Will the N*zi Party (NSDAP) hold more than 100 seats in the German Reichstag by December 31, 1930? - Will Chancellor Heinrich Brüning's government survive without collapse or new elections by December 31, 1930, following the Reichstag vote? - Will any major European democratic leader (e.g., Ramsay MacDonald, Tardieu, or Arthur Henderson) publicly warn of 'democratic erosion' in Germany by December 31, 1930, in response to N*zi gains? quoting : 'Just saw a N*zi rally in Berlin—crowds shouting about jobs and blaming Versailles. Unemployment lines are longer than ever. How many seats will they get in the Reichstag?' wtf: top influence score: Himoto. some random guy. 17 points from 2 posts. brüning, the actual chancellor: 12. tied with gandhi. street-level posting outweighed elite warnings 1.4x. that's how 1930 actually worked. fun sentiment shift : opening: 30.4% bearish, 17.4% bullish. by round 14: 27.3% bullish, 22.7% bearish. +9.9% net swing toward fascism in 15 rounds of simulated discourse. the herd flipped months before september. result: 75% YES on n*zis crossing 100 seats. real life? they got 107. talkie nailed it without ever seeing what came next. so idk if LLMs in 1930 would have stopped hitler. maybe they would've priced him in faster.

Announcing Talkie: a new, open-weight historical LLM! We trained and finetuned a 13B model on a newly-curated dataset of only pre-1930 data. Try it below! with @AlecRad and @status_effects 🧵