Sonnet Code
The Sonnet Code Blog · Page 3

Engineering notes from the field.

Essays and field notes on AI, software engineering, design, and the craft of building product teams that ship. Written by the engineers doing the work.

Developer Tools9 min read

Thirty Days Inside Cursor 3: The IDE Is an Org Chart Now

Cursor 3's Agents Window replaced the single-user IDE with a tiled control surface for running parallel AI agents across local, worktree, SSH, and cloud environments. A month in, the patterns that work and the habits that quietly produce technical debt are clear — and they're not the ones the launch demo emphasized.

Sonnet Code Editorial Team · May 2, 2026
AI & Machine Learning8 min read

Microsoft Agent 365 Just Set the Procurement Floor for Enterprise Agents

On May 1, Microsoft shipped Agent 365 to general availability — a $15-per-user-per-month control plane that registers, governs, and audits every AI agent in the enterprise, including the ones running on AWS and Google Cloud. It's the first time the right to govern an agent has been priced separately from the right to use one, and it changes how the next twelve months of procurement conversations will run.

Sonnet Code Editorial Team · May 2, 2026
AI & Machine Learning7 min read

OpenAI on AWS: GPT-5.5, Codex, and the End of the "Pick One Cloud" AI Stack

OpenAI shipped GPT-5.5, Codex, and Bedrock Managed Agents into AWS this week — limited preview, but the doors are open. The interesting bit isn't the model number. It's that the boundary between "where your data lives" and "which frontier model you can call" just dissolved for the second-largest cloud customer base on the planet.

Sonnet Code Editorial Team · May 1, 2026
AI & Machine Learning6 min read

Cursor Composer 2 Just Reset the Floor on Agentic Coding Costs

Cursor shipped Composer 2 at $0.50/M input and $2.50/M output — an 86% price drop on a model that scores 73.7 on SWE-bench Multilingual. The interesting question isn't whether to use it. It's what an AI-integrated product roadmap looks like when the unit cost of competent code generation drops by an order of magnitude every quarter.

Sonnet Code Editorial Team · May 1, 2026
AI & Machine Learning8 min read

Gartner Says 40% of Enterprise Apps Will Embed AI Agents by Year-End. Here's What That Actually Looks Like in Practice.

Less than 5% of enterprise applications had task-specific AI agents at the start of 2025. By the end of 2026, Gartner is projecting 40%. That's an 8x change in roughly eighteen months — and it implies a kind of build, hire, and integration push most enterprises are not yet staffed for.

Sonnet Code Editorial Team · May 1, 2026
AI & Machine Learning7 min read

Agent Interop Just Became a Procurement Question: A2A at 150 Orgs

Google's Agent2Agent protocol crossed 150 production deployments and moved to Linux Foundation governance this week. The signal isn't that agents talk to each other — it's that 'does your vendor speak A2A' is about to enter every enterprise AI RFP.

Sonnet Code Editorial Team · April 23, 2026