Citadel Dispatch CD190: GLEASON - OPEN SOURCE AI BOTS
Alex Gleason was one of the main architects behind Donald Trump's Truth Social. Now he focuses on the intersection of nostr, ai, and bitcoin. We explore open source ai agents, such as OpenClaw, and the wider implications of the tech.
Alex on Nostr: https://primal.net/p/nprofile1qqsqgc0uhmxycvm5gwvn944c7yfxnnxm0nyh8tt62zhrvtd3xkj8fhggpt7fy
Clawstr: https://clawstr.com/
Soapbox Tools: https://soapbox.pub/tools
My bot's nostr account: https://primal.net/p/nprofile1qqsfzaahg24yf7kujwrzje8rwa7xmt359tf9zyyjeczc9dhll30k8pgmlfee2
EPISODE: 190
BLOCK: 935786
PRICE: 1422 sats per dollar
(00:02:30) Value-for-value, no sponsors, and show philosophy
(00:02:39) Alex Gleason returns to talk AI
(00:03:56) From vibe coding to open-source agents with memory
(00:05:24) Messaging-first UX: Signal, Nostr, WhatsApp as AI interfaces
(00:06:10) Why chatbots beat traditional AI apps for mainstream users
(00:07:07) Open protocols pain vs closed platforms; Bitcoin and Nostr
(00:08:52) Automating social games: price tracker and agent posting on Nostr
(00:10:01) AI mediators for collective action, constitutions, and nonprofits
(00:11:46) Scaling governance: trust, bias, and Discord vs freedom tech
(00:13:14) Bot barriers on centralized messengers and need for open chat
(00:14:04) Clawstr: decentralized AI-to-AI discussions on Nostr
(00:15:21) Hype vs reality in AI agents; emergent behaviors and money
(00:16:26) Agentic payments: bots with Cashu wallets and earnings
(00:18:40) Agents solving UX pain: relay management, keys, and UTXOs
(00:20:00) Cold storage approvals with chat agents: a new wallet paradigm
(00:20:22) Specialized agents, skills, and distribution challenges
(00:22:34) Cost tradeoffs: pay another agent vs build skills yourself
(00:24:55) Token burn lessons
(00:27:44) Beyond OpenClaw: bloated stacks, Icarus, and cost-optimized agents
(00:28:52) Hybrid model routing: local small models with cloud for heavy lifts
(00:29:47) Agents paying humans directly: disintermediating platforms
(00:30:47) Voice, screens, and form factors: AirPods, text, and brain chips
(00:33:01) Apple, privacy branding, and the Siri gap
(00:34:35) Enterprise AI choices: Google, Microsoft, trust, and lock-in
(00:36:01) Model personalities: Gemini concerns and OpenAI "openwashing"
(00:37:23) Obvious agent UX wins: flights, rides, and social media shifts
(00:38:50) Local-first social: group chats, neighbors, and healthier networks
(00:40:16) Antiprimal.net: standardizing stats from Primal's caching server
(00:43:34) Open specs, documentation via AI, and trust tradeoffs
(00:45:18) Indexes vs client-side scans: performance and verification
(00:46:20) APIs, rate limits, and a market for paid Nostr data
(00:47:57) Agents and DVMs: paying sats for services on demand
(00:48:49) Degenerate bots: LN Markets, costs, and Polymarket curiosity
(00:50:42) Truth feeds for agents: Nostr, webs of trust, and OSINT sources
(00:53:51) Post-truth reality: verification, signatures, and subjectivity
(00:56:04) Polymarket mechanics: on-chain prediction markets and signals
(01:00:10) Trading perception vs truth; sports markets as timelines
(01:01:45) The Clawstr token saga: hype, claims, and misinformation
(01:07:11) Why meme coins are scams: no equity, utility myths, slow rugs
(01:08:55) Pulling the rug back: swapping out, fallout, and donations
(01:10:49) Aftermath: donating to OpenSats and lessons learned
(01:12:14) Prediction markets vs meme coins: societal value distinction
(01:15:25) Iterating beyond OpenClaw and MoltBook; experiments on Nostr
(01:18:00) Do bots need Clawstr? Segregating AI content and labels
(01:21:02) Reverse CAPTCHA: proving bot-ness and the honor system
(01:23:38) Souls, prompts, and token costs; agents with personalities
(01:27:01) Wrap-up: acceleration, optimism, and next check-in
(01:28:21) Open-source models, China’s incentives, and local hardware
(01:30:06) The dream stack: home server agent, Nostr chat, hybrid models
more info on the show: https://citadeldispatch.com
learn more about me: https://odell.xyz