The Good Stuff 27: Lessons Learned with AI Agents

The Good Stuff, with Pete and Andy - Episode 27: Lessons LearnedHosts: Pete and AndyEpisode Overview: Pete and Andy reflect on lessons learned from months of experimentation with AI agents, coding tools, and building software. They discuss the shift from "vibe coding" to more structured approaches, the importance of shipping usable tools, and their plans for a multiplayer version of Wingman.Key Discussion Points:Side Quests and Experimentation (01:16-08:30)Andy builds a habit tracker that evolved into a doom-scrolling prevention appPete experiments with media over QUIC and real-time streaming protocolsThe power of AI to remove gatekeeping from learning new technologiesBuilding an Nostr-based virtual pub with spatial audioVibe Coding vs. Slow Coding (09:38-16:13)Insights from working with serious engineers on AI-assisted developmentEveryone uses AI differently - no single "vibe coding" workflowThe importance of understanding your codebase architectureMoving slower to go faster: maintaining intuition while delegating implementationBuild vs. Buy: The Shopify Question (14:25-25:06)Andy's journey building an e-commerce site from scratch instead of using ShopifyThe value of understanding how things work vs. convenience of platformsLocalization of software development - kids will build these things nativelySelf-reliance as a valuable use of AI-gifted timeShipping Tools People Can Use (25:06-33:00)The critical lesson: put working demos in users' handsPlans for multiplayer Wingman to lower barriers to experimentationDesigning the business into Wingman - mapping workflows and agentsTesting at Bush Bash with live coding sessionsOrchestrators vs. Deterministic Processes (33:00-38:52)Why probabilistic orchestrator agents often fail in productionThe case for simple, deterministic workflow rulesLeft curve vs. mid curve: sometimes simpler is betterHumans should still design the business processesRate Limits and Model Selection (38:52-42:43)Claude Haiku as a solution to usage limitsRunning agents via API for unlimited usageMultiplayer mode for sharing subscriptions efficientlyThe challenge of making complex technology accessibleSimplifying the Message (42:43-48:31)Beacon demo: focus on the "moment of magic" not the complexity"Solvatur Ambulando" - solve it by walking aroundWingman's unique value: anywhere access + multiplayer agentsDon't let ego get in the way of clear communicationAI Agents Playing Games (48:31-59:09)Using game environments to test model performance for business applicationsGames as sandboxes for learning resource allocation and strategic thinkingBeyond single-agent approaches: teams of specialized agentsGeneral Catalyst's investment in gaming arenas for model testingMultiple Minds Per Task (55:13-01:01:38)Humans have multiple personalities for different contextsAgents may need similar specialization to avoid being overwhelmedFile-based handoffs between agents as a clean interfaceThe power of forcing agents to document their reasoning"Mid curve me is just like 'oh I've been so clever' - but that's not for the person on the other end that wants to look at it."