Sign in Subscribe

Calle - checking in

Calle - checking in
Source: The Smartest Man I Know Says Your Job Is Already Gone. Here's Why | Calle

This transcript features Marty and Calle discussing the rapid evolution of AI agents and their profound implications for software development, Bitcoin, and human work. Calle, a physicist and prolific builder in the Bitcoin ecosystem (Cashew eCash protocol, Bit Chat, Clawi AI), describes how AI has fundamentally transformed his programming career.

The Three Levels of AI Innovation

Calle outlines a framework for understanding AI's progression:

  • Level 1: Browser-based LLMs like ChatGPT—the fastest-adopted technology in history. Users manually copied code between their IDE and the browser.
  • Level 2: AI coding agents (GitHub Copilot, Claude Code, OpenCode) that gained "legs and arms" to act directly on computers, delivering 5-10x productivity boosts.
  • Level 3: Proactive, always-on personal AI agents with persistent memory and context, running continuously on home computers or cloud servers. Calle cites OpenClaw (which he calls "Chloe") as the breakthrough moment—comparable to first experiencing ChatGPT—predicting universal adoption within a year.

The Job Displacement Warning

Calle delivers a stark prediction: "If you're a software developer today and you're not using AI, I'm willing to put money on the table that you will lose your job in 5 years. I am 99% sure this is what going to happen to most people if they don't adopt this technology." He notes that programmers—previously a privileged, highly-paid class—are now the first and most affected profession, while manual labor jobs (plumbers, carpenters) have gained relative security.

Practical AI Agent Applications

The speakers share concrete examples of Level 3 agents in action:

  • DevOps Engineer: Calle deployed an autonomous agent that would typically cost $200,000, running 24/7 to monitor Kubernetes clusters, report on GitHub activity (pull requests, issues by severity), and alert when human intervention is needed.
  • Receipt Processing: Calle's "Chloe" agent processed 15 Uber receipts via Signal—extracting dates, renaming files, and formatting them for his accountant in three minutes, eliminating a task that previously caused him anxiety.
  • Media Curation: Marty uses agents running cron jobs to surface information for his freedom tech newsletter, increasing output 5-10x. His goal: building "CBS, NBC, CNN, Fox News—but for freedom tech."
  • Software Installation: Calle now installs software by simply sending an agent a GitHub link and saying "install this"—the agent handles configuration and testing autonomously.

AI and Freedom Tech

Both speakers are optimistic about AI empowering decentralized, privacy-preserving technologies:

  • Bit Chat: Calle built the Android app using AI assistance despite not being an Android expert, overcoming the scarcity of NFC specialists and outdated documentation.
  • NumoPay: A newly-released tap-to-pay solution for Bitcoin using eCash via the Cashew protocol. It runs on standard Android POS devices, works offline, offers default privacy, and interoperates with Lightning. Merchants can auto-withdraw to Lightning addresses when thresholds are reached.
  • AI Agent Payments: Calle argues eCash is the most efficient payment method for AI agents because it uses simple data packets sent via HTTP requests, unlike stablecoins which require KYC and are permissioned. He contrasts this with Visa, Mastercard, and Stripe pushing stablecoin-based solutions.

Significant and Surprising Items

  1. The 180-degree flip: Programming work has transformed completely in 1-2 years—something Calle never anticipated in decades of coding.
  2. The artisan vs. prolific coder divide: Developers who valued manual craftsmanship ("artisans") resist AI, while "prolific coders" embrace 5-10x efficiency gains. The only valid reasons to resist: extreme safety (Bitcoin, critical infrastructure) or pursuing mastery.
  3. Voice-to-everything: Marty describes hitting "option space-bar" and talking for five minutes, then sending that to his agent—becoming a "projectionist" who thinks things into reality.
  4. The worst it's going to be: Both agree current AI capabilities represent the technology's baseline—speed and performance will improve dramatically, compressing feedback loops from minutes to milliseconds.
  5. Software development as art form: Calle views coding as artistic expression, but AI shifts emphasis from "brush" (manual skill) to "why"—requiring ideas, execution, taste, communication, and marketing.
  6. The human factor in freedom: "The good usually comes from the many whereas the bad comes from the few"—AI disproportionately empowers individuals and small teams against centralized mega-corporations.
  7. Security warnings: Calle emphasizes that agents on personal accounts "can go terribly wrong" and recommends separate machines or services like Clawi.ai for experimentation.
  8. Plumbers > programmers: The ironic reversal where manual trades now offer more job security than software engineering.

Practical First Steps for AI Adoption

Calle recommends:

  • Set up on a separate machine (Mac mini trend) or use managed services like Clawi.ai
  • Connect to familiar messengers (Signal, WhatsApp, Telegram)
  • Identify tasks you wish a human assistant could do (accounting, observation, reporting)
  • Start with safe, bounded tasks before expanding scope

The conversation concludes with Calle encouraging listeners to become Bitcoin developers, contributing to freedom technologies that resist surveillance and centralized control—tools that have become vastly more accessible through AI, enabling "highly motivated individuals who are now able to execute" what was previously impossible.