2025 hack: "ask the AI how to prompt the AI" 2026 hack: at the end of every session, tell your agent "write down what you learned so next time you don't need me" it saves it as a skill, and reads it before the next task I now have one for rendering videos .. one for reading my Sentry errors .. one for my posting queue .. and one that writes like me each week, the agent needs me a little less not sure what I'll do soon 👀

2025–2026, imo, is the period that showed the clearest value of tokens backed by real revenue and real value capture. Thousands of projects disappeared when attention and liquidity moved elsewhere. But protocols with real products, real users and real revenue are still here. And more importantly, some of them are finding ways to return that revenue to the token. Look at the current numbers: – @HyperliquidX $HYPE: ~$60M holder revenue in 30D, with most trading fees flowing into HYPE buybacks – @CantonNetwork $CC: ~$49M in 30D, with network fees used to burn CC – @trondao $TRX: ~$24M in 30D, with network fees continuously burning TRX – @Pumpfun $PUMP: ~$24M returned to holders in 30D through token buybacks – @uniswap $UNI: ~$16M in 30D, with protocol fees now flowing into UNI buyback/burn – @ponsdotfamily $PONS: ~$15M in 30D, with a large part of revenue used for buyback and burn – @aeroxyz $AERO: ~$14M in 30D, with trading fees distributed to veAERO voters – @LaunchOnSF $STONK: ~$10M in holder revenue, mainly through market buybacks – @PancakeSwap $CAKE: ~$5M in 30D, with revenue from multiple products used to buy back and burn CAKE – @Aster_DEX aster-2:native: ~$4.6M in 30D, with most platform fees currently used to buy back ASTER For me, the more interesting model is: Real users → real fees → real revenue → real token capture. I think this will become one of the metrics worth watching much more closely in the next phase of the market. NFA.