AI News · The field notes
The Daily AI Drop - Morning Edition, November 11, 2026
OpenAI's DevDay 2026 sessions land in bulk, Anthropic's IPO filing hints cheap AI subscriptions won't last, and Qwen4 Max gets a hands-on workout.
OpenAI's DevDay 2026 sessions land in bulk, Anthropic's IPO filing hints cheap AI subscriptions won't last, and Qwen4 Max gets a hands-on workout.
OpenAI — Sites and the Next Generation of Web Apps | DevDay 2026
Turn an idea into an interactive web experience with Astra, ChatGPT Sites, and WebMCP. The session shows how to build new kinds of interfaces alongside colleagues and agents in the same place. It matters because it points to a web where the chatbot hands you a working app instead of a wall of text.
OpenAI — What Will You Build With Coding Agents That Don't Quit? | DevDay 2026
A Codex run spent over 1,000 hours keeping builds and tests passing. The session covers choosing recurring tasks, improving feedback, managing token budgets, and deciding when engineers should step in. It matters because it is the honest answer to how far an agent can run unsupervised — not a demo, but a colleague that never sleeps.
OpenAI — Harness-as-a-Service: Building with the Agents API | DevDay 2026
Build with the Codex harness through the Agents API, running agents in your own environment and keeping long-running tasks going. The key clarity is which infrastructure OpenAI manages and which parts are yours. That boundary decides whether an agent project is a weekend experiment or a production system.
OpenAI — Codex Security: From Vulnerability to Verified Fix | DevDay 2026
Find and fix vulnerabilities with OpenAI's cyber models and open-source CLI and SDK — scanning large codebases, catching issues in pull requests, and verifying fixes. It draws on OpenAI's own security work, so it is how they defend themselves, not theory.
OpenAI — What Comes After Loops?: Deep Dive Into Agentic Architecture | DevDay 2026
Traces the evolution from simple agent loops to collaborating agents, exploring the design choices and model advances behind that shift. It matters if you have felt single-agent loops hit a ceiling — this session names what comes next.
OpenAI — Building with GPT-Live and Image Generation | DevDay 2026
GPT-Live brainstorms, plays games, and creates images while coordinating work in the background. The session teaches how to add voice to agents and design for delegation, latency, interruptions, and changing requests — the practical realities of voice agents.
OpenAI — Build What's Next on the ChatGPT Platform with Plugins | DevDay 2026
Bring your tools, knowledge, and interfaces into ChatGPT and Codex with plugin extensions. Live demos show how to turn an idea into a useful, shareable experience and reach new users. Distribution inside ChatGPT is the story here.
aish_reganti — Anthropic's IPO Filing Reveals Why Cheap AI Subscriptions Won't Last
Anthropic's IPO filing offers a glimpse into the economics of large-scale model deployment: flat fees likely can't cover heavy agent usage costs. It matters because the twenty-dollar plan that runs agents for hours may not survive the math.
fahdmirza — Qwen4 Max: Hands-On Tests You Haven't Seen Yet
Thorough hands-on testing of Qwen4 Max, including a Botox injection simulation and a street food day simulation. Creative, unusual tests reveal more about a model than a benchmark table, and show where the open frontier stands against the closed labs.
mehulmpt — LLM KV Cache Explained For Beginners
Explains the KV cache from scratch: why it makes long conversations so expensive, and how EdgeRock-4B from Rock Matrix keeps it at a fixed size no matter how long the conversation gets. The clearest explanation of why long chat sessions slow down and cost more — and a genuinely clever fixed-size trick.
matthew_berman — What software provides when AI bypasses the interface
Asks the big design question of the moment: if agents can just do the task, what is the app even for? As AI bypasses interfaces, the role of software itself is up for redefinition.
matthew_berman — Gemini Carbon, NVIDIA CEO AFFAIR, Fable 5.2, Meta AI Ember, GPT Image 2.6, Minimax H3: AI NEWS
The weekly news roundup: Gemini Carbon, the Nvidia CEO affair story, Fable 5.2, Meta AI Ember, GPT Image 2.6, and Minimax H3. Your catch-up in one sitting if you missed anything this week.
pvergadia — The WORST Way to Give Feedback (And How Great Leaders Do It!)
The biggest feedback mistake is waiting too long to address an issue. Effective feedback should be immediate, accurate, direct, and kind. Not an AI story — but if you manage the humans who manage the agents, it may be the most useful ten minutes of your morning.
Listen: Listen to this episode
Subscribe: Listen to this episode