1. 📝 SUMMARY This week's call covered a wide range of topics including Patrick Chouinard's enterprise Claude rollout to roughly two thousand users, Daniel Zivkovic's low-code AI search implementation for real estate, and ongoing discussions about how AI consultants can stay valuable as vibe coding becomes ubiquitous. Members shared project updates spanning video generation, digital signage, cemetery management, and conversational commerce, while a deep dive into system prompt engineering revealed immediate ways to cut token costs and improve model focus. 2. 💡 KEY INSIGHTS Appending instructions directly to Claude Code's system prompt keeps them persistently fresh across turns, unlike CLAUDE.md content that gets buried deep in context. Patrick adopted IndieDevDan's verbosity-fix approach including an alias table and reference numbering, which dropped token consumption significantly while maintaining output quality. Daniel demonstrated that low-code AI tools like Algolia let developers sit side-by-side with business owners to tune prompts in real time, avoiding the months-long lag of isolated RAG development. When scaling a hands-on service business, staying in the field preserves raw user feedback, but bringing in employees often surfaces UX issues that founders unconsciously work around. Morgan shared that reframing business as an Infinite Game rather than a finite project reduces anxiety about never finishing, because operations are a continuous cycle of renegotiation rather than a fixed end state. Paul emphasized that coding agents need a clear roadmap and a chief of staff orchestrator to avoid open-ended scope creep; without deliverable discipline, the same tools produce endless expansion. Vibe-coded demos can look finished quickly but often collapse at edge cases and security boundaries. The gap between writing code and actual software engineering remains where professional consultants add durable value. A durable consulting niche exists with companies that have budget and need but explicitly refuse to become technical organizations, such as those in construction or hospitality. Positioning as an implementation and education layer beats competing on raw coding speed.

AI Developer Accelerator — Coaching Call - August 18 VIEW RECORDING - 115 mins (No highlights) Meeting Purpose A coaching call for AI developers to share project updates and discuss technical challenges. Key Takeaways - Enterprise AI Strategy: Patrick is deploying Claude to 2,000 users via AWS Bedrock. His strategy focuses on using skills to enforce an approved tech stack and building a versioned corporate memory system on Git. - Agentic Frameworks: For a conversational e-commerce agent, Algolia Agent Studio was recommended for its low-code approach, strong CRM integration, and enterprise focus. - Prompt Engineering: IndieDevDan's "Fixing Smartass Opus 5" technique was highlighted for its effectiveness in reducing Claude's verbosity and token usage by appending a concise instruction file to the system prompt. - Hardware & Moats: Raspberry Pi 5s are being used for standalone digital signs, but their high cost (~$250) is a barrier. A key strategy for success is building a "moat"—a unique feature like compliance with a specific state law (GRAMA)—to differentiate a product. Topics Enterprise AI Deployment (Patrick) - Deploying Claude to ~2,000 users via AWS Bedrock. - Rationale for AWS Bedrock: - Anthropic was unresponsive to enterprise inquiries. - AWS provided immediate support and acts as a reseller. - Enterprise support contracts are a non-negotiable requirement for corporate tools.

Turns out sarcasm is now a legitimate technical architecture decision—Patrick discovered that giving Claude Opus 5 a snarky personality shaves hours off verbose outputs, and Morgan is staring down a four-day backend rewrite because Convex's real-time sync is just that tempting. If you missed the chief-of-staff agent deep dive, you missed the moment we realized the "assistant" model completely collapses at 2,000 licenses. 📞 HOW THE CALLS WORK The calls can run 2+ hours. We want to make sure we're respecting everyone's time. Especially those of you who actually show up. Here's the structure: 👉 Reply to this post with your questions before the call 👉 If you submit a question and you're on the call, you go first 👉 We work through questions in the order they came in 👉 Then we open it up for everyone else If you can't make the call but want your question answered, drop it in the comments. We'll get to it. But priority goes to people who are there. The goal is simple: if you're taking the time to show up, you shouldn't have to wait behind questions from people who aren't even on the call. Morgan is weighing whether to pull the trigger on that Convex migration before onboarding new schools, Patrick is cooking up separate Claude style templates to keep the chief of staff from drowning the implementer in chatter, and Scott will be back from holiday soon to demo his Mac coding workflow replacement. If any of those threads resonate—or if you're sitting on your own AI architecture puzzle—bring it to the call. 🔗 ZOOM LINK (save this) https://us06web.zoom.us/j/81995207847?pwd=Xe6u6LmIQOmCP5VTnOwWYjDBfZNKGB.1 📅 WHEN Tuesday August 18th at 6PM ET Looking forward to seeing you on the call!

📝 SUMMARY This week's call centered on the evolution from single AI assistants to "chief of staff" multi-agent architectures, with multiple members converging on similar two-tier systems where a coordinator agent delegates to an implementer via shared repositories. Patrick Chouinard detailed his enterprise rollout of 2000+ Claude licenses and the governance patterns required, while others shared strategies for managing model verbosity, measuring real ROI versus vanity metrics, and overcoming organizational resistance to AI coding. Additional discussions covered database migrations for real-time apps, personal assistant tooling, and enterprise adoption tactics. 💡 KEY INSIGHTS The "assistant" model collapses at scale. Patrick described how coordinating many Claude assistants eventually consumes more time than it saves, leading his team to create a two-tier architecture: Claude Cowork acts as "chief of staff" coordinating with Claude Code as "implementer" through a shared local Git ledger, reducing humans to pure decision-makers. Two agents talking directly is the real unlock. Patrick and Paul Miller independently arrived at architectures where coordinator and executor agents communicate and write status to each other without human message relay, freeing the human to act only as a decision-maker. Surface decisions and wins, not just noise. Ty Wells noted that showing only problems creates a demoralizing interface. Effective chief-of-staff agents should highlight progress and successes alongside escalations, not just raw agent output. Set explicit time budgets. Paul Miller emphasized that agent tasks without deadlines can silently balloon from one hour to four, cascading delays. Decisive time constraints prevent runaway context windows. Opus 5 requires verbosity management. Multiple attendees flagged Opus 5 as unusually chatty. Mitigations include Matt Pocock's "wait, what?" skill to detect confusing output and Patrick's discovery that sarcastic, personality-driven prompts naturally produce shorter, clearer responses.

AI Developer Accelerator — Coaching Call - August 11 VIEW RECORDING - 96 mins (No highlights) Meeting Purpose To share progress, discuss AI agent architectures, and strategize on enterprise adoption. Key Takeaways - Chief of Staff (CoS) Architecture: Multiple members are adopting a CoS model, shifting from reactive assistant use to a proactive coordinator that manages tasks, prioritizes work, and initiates agent sessions, freeing up human time for high-level decisions. - Enterprise Adoption Strategy: Overcoming developer skepticism requires proving ROI with direct performance comparisons (e.g., agentic vs. traditional dev delivery) and enforcing governance via mandatory skills and plugins, not just rules. - Convex.dev for Real-time Apps: Convex.dev is being considered for a carpool app to replace Supabase. Its native real-time sync simplifies architecture, but indexing queries is critical to manage costs and prevent bill shock. - New Business Models: AI enables new models like Juan's low-cost ($1.5k) AI photo booth, which may pivot from VC funding to a distribution partnership with event venues. Topics AI Agent Architectures - Problem: Managing multiple AI assistants becomes a bottleneck, consuming more time coordinating agents than doing the work itself. - Solution → Chief of Staff (CoS) Model: