Deedy Hosts Private Event on Agentic Engineering in San Francisco
Deedy hosted a private event on Agentic Engineering in San Francisco, sharing three key takeaways: Steipete mandated OpenClaw contributors to use skill push prompt history to avoid bad PR; Trq212 utilized Claude as a video editor and interactive teaching tool for color grading; George Pickett planned ahead and guided Codex using Ousterhout principles.
The event had about 30 participants, including notable figures like Theo and Gergely Orosz, co-hosted by Tim Shi and Greg Kamradt.
In terms of market mechanisms, AI agent engineers and developers became the main participants, with event-driven attention focusing on Agentic Engineering practices, benefiting Claude and Codex users, while traditional coding processes faced pressure.
Source: Public Information
ABAB AI Insight
Deedy has been actively organizing AI engineering events, and this sharing continues its promotion of optimizing agent workflows. Earlier similar skills and pre-planning reflect the path of Agentic Engineering from experimentation to engineering.
In terms of capital pathways, participants exchanged cutting-edge practices, with strategic motives aimed at accelerating community knowledge dissemination, shifting resources from individual experiments to collective skill building and video editing tool delivery.
Similar to other private gatherings in AI engineering, Agentic Engineering is currently in a rapid iteration phase, with the three key takeaways highlighting the importance of prompt history, unknown unknowns, and pre-planning.
Essentially, this represents a technological substitution, where agent skills and pre-planning replace traditional PR and manual editing, aiming to reduce noise and enhance efficiency, leading to pricing power concentrating among engineers who master Agentic tools and driving the AI engineering industry chain towards engineering reconstruction.
ABAB News · Cognitive Law
Agentic Efficiency = Skill Usage × Pre-planning × Unknown Exploration
Traditional PR sells code volume, while Agentic engineering sells signal quality; those who enforce skills avoid 10,000 lines of garbage.
The more events, the deeper the insights; counterintuitively, private events accelerate the concentration of Agentic Engineering capital and practice.