Joe Heitzeberg • Founder at AI Tinkerers • ⏱️ 1 min read
Creating space for leading builders to share ideas, grow, and make an impact.
This week’s AI Tinkerers submissions offer a look at how we’re building more robust and capable AI systems. We saw systems tackling complex coordination challenges, like Sean Erikson’s Cloud Nirvana: Multi-Agent Patterns with 11 agents managing an organization, and Jack Jin’s OpenClaw: 24/7 Development Agent for continuous delivery.
We also saw concrete examples of AI applied to software development and problem-solving. Robb Winkle’s Reverse Ralph Loops uses LLMs for reverse-engineering, and Enrico’s Manufact Vibe: Build MCP Apps streamlines app development.
Enrico Toniato from Manufact (YC S25) presented “Build MCP apps the easy way,” showing how mcp-use and the Manufact platform help developers spin up an MCP-compliant server and interactive app via a CLI or Vibe coding in minutes. The demo walks through creating the project, wiring tool use for LLMs, and then submitting the resulting app to the OpenAI marketplace. It felt especially relevant to the community because it lowered the setup friction for standardized agent tool interfaces (people loved it). For builders, it made the next step from prototype to shareable product feel concrete.
Robb Winkle showed Reverse Ralph Loops, a FastAPI app that runs a “reverse Ralph Wiggum loop” over an existing codebase to generate a clean room specification for features or whole apps. It uses a looped LLM workflow to build a full spec, then iterates again to turn that spec into a deep research CLI. Coming from Jakib.ai and with his systems architecture background, Robb kept it practical. We liked how the loop-based approach felt immediately reusable (folks seemed to enjoy that it’s not just theory), and it fit the current shift from chat to autonomous, documented engineering.
SatisfAI revolutionizes feedback loops by deploying autonomous voice agents that conduct natural, real-time phone surveys to transform low-response customer data into actionable, venture-ready insights. Led by LumiereAI founders Imad Ez and Achraf Ez, a powerhouse duo of data scientists and full-stack engineers with deep expertise in machine learning and real-time systems.
This orchestration agent leverages structured LLM reasoning and the MuleRun execution environment to transform fragmented startup signals into a singular "next best action" and practical execution path for founders. Led by Abdoulaye Doucoure, a Conversational AI Consultant and Paris Dauphine data science alumnus with a decade of experience in management science and analytics.
Spaggl is a "SaaS factory" that uses a Neo4j-backed persistent graph memory and an orchestrated 15-step Inngest pipeline to transform single-idea descriptions into fully deployed, revenue-ready applications with synchronized UI and logic. Led by Jean-Michel Alandou, a Master of AI and veteran CTO with over a decade of full-stack experience, this project bridges the gap between experimental AI prototypes and scalable venture-ready products.
Trustworthy AI Project Companion automates the PMI-CPMAI methodology by integrating ethical governance and risk assessment directly into the AI development lifecycle through a guided, documentation-generating assistant. Led by Leo Fatayri, an AI Delivery Lead with 14+ years of experience scaling enterprise SaaS and chatbot ecosystems across EMEA and the US.
You are one of 95,000+ readers from Apple, Amazon, Microsoft, Google, Nvidia, OpenAI, Anthropic, Cohere, ElevenLabs, Scale AI, Groq, Mistral AI, and others — spanning frontier labs, big tech, startups, and top universities.
🛠️
Want to be in the next issue?
Every demo here came from someone who showed up and built something.
![⚡ See 24/7 Dev Agents + MCP Apps [AI Tinkerers - Post-Training]](https://sloppy-joe-app.imgix.net/ai_tinkerers/logos/post-training_laptop-sticker_190_de2487c31e50813a3471e79627268ffb.png?w=300&fit=max&auto=format%2Ccompress&q=75){kind=small}
⚡ See 24/7 Dev Agents + MCP Apps