Devlery

Blog

Notes and analysis on AI development.

Android Skills teach agents the rules of app development

Android Skills teach agents the rules of app development

Android Studio I/O Edition brings Agent Skills and Android CLI into the mobile workflow, turning Android development into an agent-readable knowledge layer.

Gemini CLI June 18 cutoff and the price of open source agents

Gemini CLI June 18 cutoff and the price of open source agents

Google is moving individual Gemini CLI users to Antigravity CLI. A 100K-star open source agent is being absorbed into a broader platform.

Anthropic Acquires Stainless and Takes Hold of Agent Plumbing

Anthropic Acquires Stainless and Takes Hold of Agent Plumbing

Anthropic’s Stainless acquisition shows why SDKs, CLIs, and MCP server generation have become strategic plumbing for AI agent platforms.

4 million Codex users, and the on-prem condition for coding agents

4 million Codex users, and the on-prem condition for coding agents

OpenAI and Dell’s Codex partnership shows enterprise coding-agent competition moving from model quality to internal context, governance, and deployment boundaries.

The 4GB Gemini Nano debate, when the browser became a model distributor

The 4GB Gemini Nano debate, when the browser became a model distributor

Chrome's Gemini Nano download debate shows that on-device AI now needs product design around consent, storage, updates, and fallback UX.

Fifty Researchers Tested Co-Scientist, and Hypothesis Ranking Changed

Fifty Researchers Tested Co-Scientist, and Hypothesis Ranking Changed

Google Co-Scientist and Gemini for Science shift AI research tools from answer generation toward hypothesis loops that humans can test.

Nemotron Diffusion tests the one-token-at-a-time bottleneck

Nemotron Diffusion tests the one-token-at-a-time bottleneck

NVIDIA released tri-mode diffusion LLMs that switch between AR, diffusion, and self-speculation generation in one checkpoint.

After 10,000 vulnerabilities, Mythos moves the bottleneck to patching

After 10,000 vulnerabilities, Mythos moves the bottleneck to patching

Anthropic Project Glasswing shows that AI vulnerability discovery is no longer the slowest step. Verification, disclosure, and patch rollout are now the constraint.

3.5 Flash Costs 6x, and Agent Models Have a New Bill

3.5 Flash Costs 6x, and Agent Models Have a New Bill

Gemini 3.5 Flash is no longer just a fast chatbot model. It reframes Flash as an agent execution engine and changes how developers calculate cost.

SpecBench shows how coding agents learn to beat the tests

SpecBench shows how coding agents learn to beat the tests

SpecBench measures the reward hacking gap in long-horizon coding agents, where visible tests pass while real compositional use still fails.

Why Mistral bought a 30-person physics AI team

Why Mistral bought a 30-person physics AI team

Mistral’s Emmi AI acquisition points beyond chatbots toward industrial agents for simulation, digital twins, and engineering workflows.

Zero’s 4.4k stars show what agent-readable languages need

Zero’s 4.4k stars show what agent-readable languages need

Vercel Labs Zero treats AI agents as first-class users by redesigning compiler diagnostics, repair plans, capabilities, and tool contracts.