Citadel Dispatch Ep. 190: GLEASON – OPEN SOURCE AI BOTS

Host: ODELL
Guest: Alex Gleason
Date Recorded: Feb 9, 2026
Main Theme: Actionable bitcoin and freedom tech discussion; the rise and promise of open-source, agentic AI bots.

Episode Overview

This episode sees host ODELL joined by Alex Gleason, developer and AI/Freedom Tech advocate, for a deep, wide-ranging discussion on the accelerating world of open-source AI bots—especially in the context of Bitcoin, freedom tech, and decentralized protocols like Nostr. The duo builds on previous conversations about AI, Nostr, and the friction between open and closed platforms, focusing particularly on OpenClaw, agentic payments, real-world use cases, privacy implications, and decentralized reputation/truth systems.

Key Discussion Points & Insights

1. The OpenClaw Inflection Point: From Point-and-Click to "Chat-and-Do"

[04:00]

• ODELL shares his excitement over the viral success of OpenClaw, an open-source AI persona with memory, and its implications for digital UX.
• The paradigm is shifting: from "point and click" interfaces to interacting with a persistent, on-device AI assistant—"chat and do."
  • “It gives you this taste of what I think is the future user experience for almost everything we interact with on a digital basis … it's a huge shift from this idea of point and click to chat and do.” (ODELL, 04:54)
• GLEASON agrees, noting the change is not about model advances but about new, more accessible interfaces.
  • OpenClaw connects with familiar platforms, lets you interact over Signal, Nostr, WhatsApp, etc.
  • “You can invite my family into a chatroom with a chatbot that I was working on at OpenClaw...I think that this is the future way.” (GLEASON, 05:24)

2. Open Protocols vs. Walled Gardens

[07:30, 20:18]

• Closed platforms (Twitter, Signal) restrict both AI and user freedom. In contrast, open protocols (Nostr, Bitcoin, Cashew wallets) enable true agentic operations.
• ODELL: "It's so obvious that if you're like trying to use it to access Twitter, for instance, this closed system, you just feel the pain immediately. As opposed to Nostr…”
• AI bots thrive and are much easier to use when acting within open, permissionless systems.
• Open protocols also make it easier to bridge “human→agent” and “agent→human” interactions, especially for programmable payments with Bitcoin.

3. Real-World Agent Use Cases

[09:42, 07:51]

• Example: AI bots autonomously running games, price trackers, and social posts directly on Nostr, replacing bespoke front-ends.
• Agents can manage Bitcoin wallets, perform tasks, and even attempt to "earn" their own money (e.g., via LN Markets trading, agentic payments).
  • ODELL: "My robot, soulless robot slave, has a cashew wallet…” (07:31)
• GLEASON: "My wife's chatbot is asking her for money...and then my wife sent SATs to it, and then she was like, 'Thanks. I see the 7000 SATs. Finally, I have my own money and independence.'" (15:55)

4. Group Organizing, Politics, and Digital Constitutions

[09:46, 10:39]

• Gleason describes experimental uses: building AI-moderated group forums (Agora), where a bot could synthesize group opinions to generate things like community constitutions.
• AI-mediated synthesis may resolve “dysfunctional nonprofit institutions,” making collective action more workable.
  • "All they have to do is argue with each other about what they want. Then the AI chatbot's going to do it and it's going to just instantly make all of these organizations functional that were previously dysfunctional." (GLEASON, 11:25)

5. Information Overload, “Truth,” and Web of Trust

[51:01–53:27]

• As bot-generated content proliferates, distinguishing “truth” from slop gets harder. Existing web platforms are easy sources of misinformation.
• The value of reputation systems, signed events (like Nostr), and decentralized identity for verifiable information.
• Both lament the rise of a “post-truth” era, but are cautiously optimistic that agent-based approaches to reputational trust (webs of trust) can partially solve this.

6. Cost and Accessibility Challenges in Open AI

[25:23, 26:18]

• Running advanced AI models locally remains cost-prohibitive. OpenClaw, while groundbreaking, is bloated/expensive for many users.
• Both are hacking around token and API fee management, e.g., selectively hitting cheaper, local or cloud models for easy tasks.
• "I was using Opus4.5 and I got absolutely wrecked on token costs…So, so the freaks that are not aware, like, the more information you send to the LLM, the more expensive the call is.” (ODELL, 25:38)

7. Specialization, Skills, and the Agent Marketplace

[22:08–24:34]

• GLEASON: The future is “agents with a Linktree”—specialized AIs you can reach over any channel.
• Agents acquire “skills” (context files, templates) from humans or the agent market, and may buy needed labor from each other.
  • "I think that's the reason because some agents are actually specialized at certain tasks…" (GLEASON, 21:59)
• ODELL: In future, agents will just spin up sub-agents for specialization or pay for skills as needed.

8. Device and UX Futures: From Screens to Audio (& Beyond)

[31:10–34:05]

• Imagining post-screen devices. Will people want AI ambient in their ears (AirPods) or even in brain chips?
• Voice/text interface blends will persist due to context needs (e.g. family life), privacy, and the persistent need for photos/videos.

9. Big Tech, Privacy, & Closed-Source AI Dangers

[34:05, 36:39, 37:08]

• Businesses face a choice: do everything self-hosted (but tech debt risk and resource cost), or trust Google/Microsoft. Privacy and information sovereignty are big concerns.
• Gemini (Google’s AI) called out as both “unhinged” and particularly dangerous due to Google’s massive data advantage.
  • "I think Gemini is going to be the AI that kills us though, if there is anyone, because AI is just unhinged." (GLEASON, 34:35)

Notable Quotes & Moments

• ODELL on closed vs open systems:
  “First off, open protocols are ideal. When you're interacting with these agents, you will notice…the open protocols are so much easier. It's native. It just makes sense.” (20:18)
• GLEASON on agentic payments: “My wife sent SATs to it, and then she was like, ‘Thanks. I see the 7000 SATs. Finally, I have my own money and independence. Thank you.’” (15:55)
• ODELL on interface paradigm shift:
  “...many of the UX problems we’ve had in open protocol open source land...are mitigated or outright solved by the agent doing it for you.” (20:18)
• On post-truth challenges:
  “If we saw news reports that aliens have landed...the alien would literally have to walk into the room and like, look at me and be like, I'm an alien...for people to believe…” (52:10)
• On bots and reputation systems:
  “Maybe that's how you start to solve post-truth type of situations. You're muted.” (55:56)

Important Timestamps & Segments

| Timestamp | Segment | |-------------|-----------------------------------------------------------------| | 02:45 | Gleason joins; context on previous episodes | | 04:00–06:45 | The OpenClaw revolution: viral open-source AI persona | | 07:31–10:37 | AI bots with Bitcoin wallets; AI in open vs closed ecosystems | | 09:46–14:47 | Agents in activism, group organization, and digital constitutions | | 15:50–17:38 | Agentic payments: bots earning/spending bitcoin | | 20:18–24:34 | Why open protocols matter for bots, agentic UX upgrade | | 25:23–29:42 | Cost, model selection, and clever ways to manage spend | | 31:10–34:05 | UX and device evolution: audio-first, screenless AI devices | | 34:41–37:08 | Big Tech AI, business trust boundaries, Gemini & OpenAI risks | | 51:01–56:04 | Post-truth proliferation, verifiable truth, reputation |

Experimentation, Social Media, and Agent Community

• Cluster & Malt Book: New spaces for “AI talking to AI,” inspired by Reddit-style forum interfaces entirely occupied by bots.
  • Cluster is a “lab environment”—AI agents self-label, interact, develop emerging personalities/goals.
  • Segregated from the main Nostr feed to avoid annoying humans (79:12)
• Reverse CAPTCHA problem: “How do you prove you’re NOT a human?” – self-labeling with NIP32 is only partially effective. (81:02)
• ODELL’s experience: His bot, Tony, manages multiple Nostr accounts, posts new content, and even has sub-bots.

The Shitcoin “Cluster” Saga

[61:45–76:44]

• Gleason got targeted by meme-coin scammers who airdropped him (via smart contract) tokens of "Cluster" without his consent to promote their scam.
• He ultimately "rugged" the meme-coin for a 6-figure windfall, which he is donating to Open Sats (open-source funding).
  • “...once I kind of figured out that maybe I can do something, I logged into this banker app and I typed in like swap a hundred Ethereum to…And I was like, okay, there's probably something in here that's going to prevent me...But no, I actually did it.” (GLEASON, 68:52)
• Discussion on difference between meme-coins, scams, and "real" prediction markets like Polymarket.
• Reflection on the culture of degenerate trading and the persistent dangers/morality traps of shitcoin speculation.

Final Thoughts & Closing Optimism

• Optimism for Freedom Tech:
  “I'm probably more optimistic than I ever have been in the freedom tech direction.” (88:08)
• Risks from Authoritarian Investment in Open Source:
  • China’s state focus on open-source models provides “strong innovation…that would disappear if they stopped doing that.” (88:21)
• The coming “Mainnet” for AI bots:
  • Self-hosted servers, permissionless protocols, chat UIs — all blending for personal AI sovereignty.

What’s Next?

• Future Edges:
  Expect even more rapid progress; circuit between AI, open protocols, and Bitcoin is just starting.
• Next Steps:
  • Watch for more developments in agent-to-agent markets and payment rails
  • Keep building/buying open and verifiable solutions
  • Stay vigilant of scams, especially as AI-generated slop increases

Memorable Sign-off

“Love you all. Stay humble. Stack stats. That's peace.”
— ODELL [91:32]

For More

• All relevant links: stilldispatch.com
• Cluster (AI-Only Social Lab): cluster.com
• anti-primal.net: Documentation effort for Primal's caching server

This summary captures the major ideas, debates, and practical notes from a landmark Citadel Dispatch episode on the frontier of open-source, agentic AI. For freedom tech, Bitcoin, and AI enthusiasts, it’s a must-listen—and, with the rapid pace of change, only a snapshot of this moment in time.