Best of terminallyonchain.xyz on Kiwi News
Most upvoted by Kiwi News readers
14 stories from terminallyonchain.xyz have been shared on Kiwi News since 6 September 2024, with 37 upvotes and 2 comments in total. The latest was shared on 15 March 2025. They are listed below, most upvoted first.
Text under a title is an AI summary of the linked article.
- The inevitable rise of brand marketing on Zora
Terminally Onchain's author YB argues that brand accounts should start marketing through onchain mints on Zora now, even small companies, because the goal is to build an onchain distribution channel and CRM before the rest of the ecosystem catches up. YB cites large examples such as Doodles' weekly animation mints, which have reached 400k mints, and Coinbase publishing quarterly earnings reports as mints. YB reports that after about 6 weeks of roughly 25 mints, the Zora follow count rose from under 100 to 1,600, and recommends starting with raw company content, then targeting collectors with tools such as Hypersub and Boost.
- It's not a geopolitical red flag but an open source green flag
The newsletter argues that the DeepSeek R1 release is an open-source milestone rather than a geopolitical threat, and it draws on Yishan's view that the model's open weights and published paper make it comparable to Google's 2004 S-1 moment. DeepSeek, a Chinese AI startup, released R1 under the MIT License, matching or surpassing OpenAI's o1 in performance, with the model reportedly developed for under $6 million and API pricing of $0.14 per million tokens versus OpenAI's $7.50. The piece also covers Venice, an AskVenice product whose VVV token hit a $350 million market cap in a day, and notes that Coinbase listed the token on its first day, a move some community members criticized. The author flags that Venice's privacy claims are limited, since any single chat is visible to the GPU provider handling it.
- Upcoming trends in Q1 for the agent meta
The author argues that agent frameworks are not simply this cycle's L1 trade, and that comparing valuations to last cycle's L1 prices can lead to misguided price targets. Ai16z reportedly crossed a $2 billion market cap and Virtuals crossed $4 billion over the holidays, up from under $80 million and about $350 million in late October. The post also says attention on agent infrastructure may be nearing a local top, with consumer agent projects such as those focused on community management likely next, and it highlights aixbt, a Virtuals-launched trading agent, as a leading example in its vertical.
- The Availability Cascade for the Onchain Agentic Economy
YB argues that the recent crypto-AI experiments, including truth_terminal, Aethernet, Forest, Luna, and the price action of $GOAT and $HIGHER, signal an early availability cascade for an onchain agentic economy that will eventually be seen as a revolution rather than a fad. Citing Joel Spolsky's claim that good software takes 10 years, YB says DAOs are now around year 8 and that AI agents will act as the glue steering decentralized organizations and headless brands. YB compares this to the limited liability corporation, drawing on Coase's transaction cost argument and Gilbert's three ideas, and maps parallels to minimized transaction costs in DeFi, agents as artificial persons, and tokens as tradeable shares. YB says the timeline is years, not months.
- Your new mission is to join the right WhatsApp chats
As AI tools spread, the key advantage shifts from surfacing public information to holding secrets others lack. Tyler Cowen notes that humans know secrets while AIs generally do not, and Ben Thompson argues that prediction markets let people profit from secrets before they reach training data. YB says the $200-a-month OpenAI deep research subscription is out of reach for most people but expects its capabilities to reach cheaper plans by the end of 2025.
- Terminally Onchain but Artificially UNintelligent
YB, writing in the Terminally Onchain newsletter, argues that agents, the current focus of onchain crypto-AI projects, are a short-lived phase and that the wider Onchain AI stack, including middleware, compute, and data, will draw more attention next. He says agents are likely to hit a local top and see consolidation, with a few projects such as aixbt and zerebro continuing development while infrastructure teams like ai16z and Virtuals refine their approaches. He predicts that data tokens will become the next major asset class and plans to cover decentralized AI projects such as Gensyn, Olas, Vana, and Bittensor in 2025.
- The Dark Knight of Consumer Crypto
The author argues that every consumer crypto founder should consider zkTLS web proofs for their product and distribution, since they let users prove facts about their web2 data without revealing personal information. The article says APIs were controlled by platforms that later restricted access, such as Twitter in 2012, Facebook in 2014, and Reddit's 2023 API price change, which made it hard to compete. Web proofs are described as "unruggable APIs" that a website cannot stop, and examples include proving a bank balance of at least $5000 or Spotify streaming activity. The article names Reclaim Protocol and Opacity as the two main players in the space.
- From Binary to Analog Prediction Markets
TMR.NEWS launches as what its creators, Small Brain and Sofiane, call the first natural language prediction market, where users bet on what tomorrow's NYT front page headline will be. The article argues that the current yes/no model of prediction markets only scratches the surface, and that AI-judged, open-ended "analog" markets can ask far more meaningful questions. The product runs on Base with sub-cent bets, uses Reclaim Protocol's zkTLS to verify the NYT headline, and relies on GPT to judge which submissions are closest in meaning to the actual headline. The author also flags a weakness: NYT staff who know the headline could game the system, and identity verification may be needed as a fix.
- The Availability Cascade for the Onchain Agentic Economy
YB argues that crypto Twitter's recent AI-related excitement is an availability cascade that will grow into a lasting shift toward an onchain agentic economy, and that this could rival the impact of the limited liability corporation. He points to recent experiments such as truth_terminal, Aethernet, Forest, Luna, and the price moves of $GOAT and $HIGHER as early signs. He draws parallels between the LLC's formation and today's developments: lower transaction costs (DeFi versus TradFi), artificial persons (agents), and tradeable shares (tokens). He says DAOs need AI agents to steer them, and he frames the shift as slow, with no expected timeline of "by Christmas."
- Dropping $7.5k to source high quality alpha
The author describes spending about $7.5k to buy 10,000 $COOKIE tokens to access premium features of Cookies.fun, a crypto AI-agent analytics platform, and says the purchase was not sponsored. They say they want a faster daily routine for tracking emerging narratives, and they praise the tool's mindshare, holder and engagement dashboards. The post tracks 1,000+ agents and reports that the $COOKIE token's market cap was hovering between $150 million and $200 million, with a Binance perps listing that day. The author also notes that Virtuals integrated the Cookie DataSwarm APIs and that they spoke with the founders in December.
- Niche down or die tryin
Farcaster is shifting its channels toward "cozy corners," small invite-curated communities, instead of broad topic channels like /crypto or /books, as part of a 1.5-year experiment the team will end if the plan fails. The article says that after the June 2023 launch and a permissionless opening, channels grew to over 3500, most of them small and inactive. It argues that broad channels are too general and that most of their casts could be home-feed posts. Under the new model, hosts would manually add or invite members, as in a group chat, while hashtag-style topics would become umbrella themes for several niche channels. The author raises unresolved questions about channels such as /techarchive, /newyork and /paragraph.
- Terminally Onchain by YB | Devin, here's 50 USDC. Get it done
YB's newsletter post is a resource guide to decentralized AI, arguing that AI agents may become a major driver of stablecoin usage in the coming years. YB writes that AI development is easier to scale through developer solutions than through human adoption, which faces friction in onboarding users and merchants. The post summarizes a Cyber Fund write-up, shared by Sean Brennan of ChainAgent, on the "Internet of Intelligence," where crypto wallets, privacy protocols, decentralized marketplaces and stablecoins supply the financial layer for AI agents. It also lists open questions and bookmarks, including links to Coinbase's AI wallet documentation and NovaNet's post on USDC and AI agents.
- Grab a beer, we're going Bier mode
The author argues that crypto infrastructure has improved substantially since the 2020-21 bull market, so the space should now focus on consumer app experiments that stress-test the tooling. Citing a 2018 USV essay by Dani Grant and Nick Grossman, the post holds that apps beget infrastructure, and that the current moment is another turn of the apps-infrastructure cycle. It lists recent improvements such as smart wallets, EIP-4844 gas fee reduction, and Farcaster frames, and it praises a demo by Farcaster developer @slokh of an app where users sign in with email and pay with Apple Pay to collect an NFT. The post urges builders to create fast, low-cost "crypto hooks" and to test many simple ideas, as Nikita Bier does.
- Grab a beer, we're going Bier mode
The author argues crypto infrastructure has improved enough since the 2020-21 bull market that new consumer apps should now be built to stress test the tooling. Citing Dani Grant and Nick Grossman's 2018 USV essay "The myth of the Infrastructure Phase," the post argues that apps beget infrastructure and that crypto is in another turn of the apps-infrastructure cycle. It lists recent tools such as Coinbase and Privy smart wallets, Syndicate Cloud transaction bundling, EIP-4337 gas abstraction, and EIP-4844 gas fee reduction on Base. It holds up @slokh's Farcaster demo, where an app let users sign in with email and pay with Apple Pay, as a model, and calls for quick, simple app experiments.