Anatomy of Inter-Agentic APIs
summary
Claims an offline inference at source platform that runs air-gapped on-device and a VS Code co‑pilot called M9 developer, demonstrated generating code, fixes and tests with no internet. Describes agent topologies (star, hierarchical, mesh), SPA design priorities (security privacy adaptability), and use of small LLMs (default Llama 3.1, swappable). Cites a reflection pattern reaching ~90% of GPT‑4, reports ~150 active users, and shows the product working in a recorded offline demo.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:09Software agents overview
- 5:46Agentic architecture patterns
- 10:00Agent topologies
- 12:08Why agents matter
- 15:46API design fundamentals
- 17:52AI API semantics challenges
- 23:11Business relevance of agentic AI
- 27:08Live demo: offline AI inference
- 31:25M9 inference platform design
- 33:54M9 Developer agentic APIs
- 35:19Closing thoughts
- 36:01Q&A
related talks
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
37mEmpowering Developers with local AI
29mAPI Integration augmented by GenAI
48mBuilding Offline AI Agents in Your Mobile App
46mThe AI-Powered Developer
32mProgramming language evolution in the AI era
24mAI Agents in Production: Why the Platform Matters
29mGenAI in Backstage - Turbocharge Your Developer Productivity!