AI Developer Accelerator
Log In
Community
Classroom
Calendar
Members
Leaderboards
About
Log In
5
Patrick Chouinard
3d •
General discussion
AI Developer Accelerator — Coaching Call - September 15th
AI Developer Accelerator — Coaching Call - September 15
VIEW RECORDING - 83 mins (No highlights)
Meeting Purpose
A weekly sync on AI projects, strategies, and development tools.
Key Takeaways
Agentic Workflows Drive Efficiency: Custom agents (e.g., Patrick's "Beacon" news aggregator, Ty's "UI in reverse" code generator) are automating complex tasks, but this high-volume usage is rapidly consuming token limits.
Strategic Pivots for Revenue: Juan is shifting from development to sales for his AI photo booth, creating bespoke transformations for high-end venues to secure initial clients.
Operationalizing AI for Consulting: Rod is exploring a consulting product to help SMBs operationalize AI, focusing on creating custom skills and structured workflows to reduce token waste.
New Member Onboarding: Lisa, a full-stack engineer, joined to learn about agentic systems and is evaluating development harnesses like Hermes, T3 Stack, and PI.
Topics
Agentic Workflows & Token Management
Patrick's "Beacon" News Aggregator:
Workflow: "Beacon" (research agent) → raw report → "Clara" (contextualization agent) → TTS audio → internal archive (audio + transcript).
Goal: Create a continuous improvement loop by feeding user feedback from ChatGPT back into the system to refine future briefings.
Ty's "UI in Reverse" Development Method:
Process: Design UI mockups → agent generates code to meet UI contracts → adversarial agent tests the code.
Rationale: This method was used to build an interactive pitch app for "The Island" platform, which replaces $50k in subscriptions for a Bahamas-based company by consolidating 5-8 services.
Benefit: The app captures user analytics (clicks, shares) to track engagement, unlike a static slide deck.
Token Consumption:
High-volume agentic work is rapidly depleting token limits, even on upgraded plans.
Example: Ty's UI testing consumed all Astra tokens in two days.
Mitigation: Patrick's team reduced token use by building a custom "Grille with docs" skill with pre-defined business context, cutting spec-generation questions from 70–80 to 15–20.
Project & Business Updates
Juan's AI Photo Booth:
Strategy: Pivot from development to sales to generate revenue.
Target: High-end venues in San Diego.
Sales Funnel: Email with bespoke transformation demo → call → in-person demo (using a phone/iPad setup) → final follow-up email.
Ryan's Projects:
Estate Agency Website: A new site with a CRM backend is outranking competitors (e.g., Savills) in local SEO within weeks of launch.
Hermes Upgrade: Added telephone capability, enabling the agent to make calls.
Fitness App: Building a multi-platform app (Mac, iPhone, Watch) for personal fitness tracking, including meal logging via image analysis.
Paul's Visual Shelf Recognition Project:
Reinvention: Automating the training and fine-tuning of a 12-year-old shelf recognition project using open-source visual LLMs.
Benefit: Eliminates the need for expensive third-party vendors and manual image annotation.
Tools & Infrastructure
Lisa's Onboarding:
Background: Full-stack engineer exploring agentic systems for SMB consulting.
Goal: Build a new development environment on Fedora and evaluate agent harnesses.
Harness & Infrastructure Discussion:
Hermes: Used for orchestration and daily management; integrates with OpenRouter for model access.
T3 Code: Recommended for managing multiple Claude/OpenAI subscriptions and isolating dev/staging/prod environments.
PI: An open-source tool being explored for specialized tasks like code review.
ccblackbox.app
: Ryan is using this open-source, Mac-only IDE for Claude development.
Model Selection & Cost:
Strategy: Use cheaper models (e.g., Sonnet, GLM 5.3) for routine tasks and reserve expensive models (e.g., Opus) for complex work.
GLM 5.3: Noted as a cost-effective alternative to IQ models on OpenRouter.
Next Steps
Juan:
Create a landing page for the AI photo booth.
Develop bespoke transformations for target venues.
Execute the sales outreach plan.
Lisa:
Research Hermes, T3 Code, and PI to select a development harness.
Ryan:
Complete the multi-platform fitness app.
Patrick:
Finish the browser extension for exporting ChatGPT conversations.
Paul:
Deploy T3 Code servers to a home server for overnight processing.
Action Items
Expose Beacon Morning Briefing archive via Traefik + Tailscale -
WATCH (5 secs)
Create bespoke image-to-image style for target venue; then email venue director w/ demo + offer free pilot -
WATCH (5 secs)
Build photo booth landing page -
WATCH (5 secs)
Identify and join local country club/venue networks -
WATCH (5 secs)
Prepare portable demo kit (phone + iPad) and pre-recorded demo for venue visits -
WATCH (5 secs)
Evaluate GLM 5.3 on OpenRouter; set Claude benchmarks; run OpenRouter tests -
WATCH (5 secs)
Run ERP launch postmortem w/ team -
WATCH (5 secs)
Evaluate T3 stack, Hermes, OpenRouter; then report back to group -
WATCH (5 secs)
Deploy T3 code servers on Mac Mini + Linux server -
WATCH (5 secs)
Clean up and update personal projects portal -
WATCH (5 secs)
Like
0
0 comments
5
AI Developer Accelerator — Coaching Call - September 15th
AI Developer Accelerator
skool.com/ai-developer-accelerator
Master AI & software development to build apps and unlock new income streams. Transform ideas into profits. 💡➕🤖➕👨💻🟰💰
11.4k
Members
53
Online
8
Admins
JOIN GROUP
Leaderboard (30-day)
1
Scott Rippey
+7
2
Lucas Lim
+7
3
Ali BabaStudios
+4
4
Takeichi Inoue
+3
5
Akash Narayan
+3
See all leaderboards
Powered by