4d • General discussion
AI Developer Accelerator — Coaching Call - August 18th
AI Developer Accelerator — Coaching Call - August 18
Meeting Purpose
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)
- 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.
- Tech Stack Enforcement: Building a skill to guide non-technical users to approved internal tools, preventing recommendations for unapproved platforms (e.g., Vercel, Superbase).
- Level Zero Support: Creating a support skill from a pilot's knowledge base to automate answers to common questions, filtering Claude's responses with internal context.
- Corporate Memory System: Developing a versioned knowledge base on Git (using Markdown with Frontmatter) to manage team and corporate knowledge.
Project Updates
- An AI photo booth service using diffusion models for image transformation.
- Focus: Developing 30+ transformation styles and building an Instagram portfolio.
- Scaling Challenge: The service is hardware-dependent. Juan is hiring an "AI boot technician" to scale operations while he retains a field presence for direct user feedback.
- A digital sign system running on Raspberry Pi 4/5s.
- Core Function: AI transforms rough inputs (e.g., photos of scribbles) into polished, cohesive slides with auto-generated QR codes for events and locations.
- "Moat" Example: The Heritage Plot cemetery management system is built to comply with a specific state's Freedom of Information Act (GRAMA), a feature no competitor offers.
- Shopping List App: A custom app replacing Apple Notes, with features like ingredient scanning, dietary restriction checks, and boycott list alerts.
- No-Screen Voice Device: A project to build a voice-only device for kids to call grandparents.
- Hardware Challenge: The high cost of Raspberry Pis is a barrier. Alternatives like Arduino or Pi Zero were suggested.
- Experimenting with MiniMax H3 and Comfy UI for video creation.
- Result: Created a "fantasy cat lion" cartoon, exploring merchandising opportunities (kids' story, plush toys).
- Manages a large project using a Mac Mini for 24/7 agent operation, freeing up his laptop.
- Agent Workflow: Claude Fable (high-level) → Opus 5 (code generation) → Codex (validation).
- Scaling: Traveling to the Philippines to recruit a support team, allowing him to focus on development.
Prompt Engineering & Agent Frameworks
- Technique: Appending a concise instruction file to Claude's system prompt.
- Rationale: System prompts have higher priority than user prompts and are always fresh, ensuring instructions are consistently followed.
- Features:
- Aliases: Shortcodes (e.g., ELI → "Explain it to me like I am a 12-year-old") for efficient, reusable commands.
- References: Numbered items (e.g., R1, F2) for unambiguous reference to past context.
- Use Case: A conversational e-commerce agent for knowledge retrieval, real-time data (order status), and actions (scheduling).
- Recommendation: Algolia Agent Studio.
- Rationale: A low-code tool with strong CRM integration, ideal for e-commerce. It offers a free tier for small projects and excels at grounding answers in a knowledge corpus to prevent hallucinations.
Next Steps
- Hemal: Evaluate Algolia Agent Studio for the e-commerce conversational agent.
- Juan: Involve a non-technical assistant in the AI photo booth to gather unbiased user feedback.
- Shakur: Research Arduino or Pi Zero as lower-cost hardware alternatives for the voice device.
- Daniel: Contact Ty for a security audit of a ShipKit project.
Action Items
- Post ComfyUI MiniMax H3 workflow link in group chat - WATCH (5 secs)
- Confirm AI photo booth event this weekend - WATCH (5 secs)
- Contact social media helper re: this weekend's event; brief on capture + user feedback - WATCH (5 secs)
- Review Arduino R4 + USB keypads for no-screen messaging device - WATCH (5 secs)
- Email Ty re: ShipKit security audit; request quote + availability - WATCH (5 secs)
0
0 comments
AI Developer Accelerator — Coaching Call - August 18th
skool.com/ai-developer-accelerator
Master AI & software development to build apps and unlock new income streams. Transform ideas into profits. 💡➕🤖➕👨💻🟰💰
Leaderboard (30-day)
1
+3
2
+3
3
+2
4
+1
5
+1
Powered by