AI Coding Assistants

Today's Stories

1. 1

   Claude Code turned every engineer into three. Now companies need more product thinkers

   Claude Code turned every engineer into three. Now companies need more product thinkers

2. 2

   Using Local Coding Agents

   Using Local Coding Agents

3. 3

   The 33-year-old executive Satya Nadella is trusting to fix Microsoft’s Copilot AI assistant

   The 33-year-old executive Satya Nadella is trusting to fix Microsoft’s Copilot AI assistant

4. 4

   AAON's water-free cooling system powers Applied Digital data centers

   AAON has developed a cooling chiller system for Applied Digital's data centers that does not use water. During North Dakota's long winters, the system's compressors shut off and allow outside air to do the cooling work. Data center cooling is a major operational cost and environmental concern. A water-free approach removes a significant constraint on where large AI compute facilities can be built, potentially making regions like North Dakota more attractive for hosting the chip-intensive operations that companies like Applied Digital rely on.

   Applied Digital operates massive data centers that depend heavily on chips from Nvidia, so the efficiency of their cooling infrastructure directly affects their ability to scale operations and manage costs.

5. 5

   Linux Foundation launches Akrites to patch open-source flaws before AI exploits them

   The Linux Foundation announced Akrites, a coordinated industry initiative bringing together about twenty tech companies—including Amazon Web Services, Anthropic, Google, IBM, Microsoft, NVIDIA, and OpenAI—to fix security vulnerabilities in widely used open-source software before attackers can exploit them. At its core is a shared Security Incident Response Team that acts as a single point of contact for maintainers instead of dozens of organizations independently reporting the same flaws. AI models can now scan large open-source projects in minutes instead of weeks, exposing flaws far faster than before. This shifts the balance: attackers without deep technical skills will soon have the tools for sophisticated exploits. Currently, fewer than five percent of validated open-source vulnerabilities from recent months have been patched, and maintainers get buried under duplicate reports while real bugs get lost in noise. A shared response team aims to cut through this inefficiency.

   When a critical package no longer has an active maintainer—a common problem in volunteer-run projects—Akrites plans to step in as a "maintainer of last resort" and ship fixes itself so patches reach all users in time. Seed funding comes from Alpha-Omega, a directed fund under the Linux Foundation, and other organizations can contribute engineering resources or funding.

6. 6

   GitHub Copilot matches rivals on fewer tokens

   GitHub published benchmark results showing its Copilot agentic harness—the shared engine powering Copilot CLI, the Copilot app, and code review—achieves task-completion rates on par with Claude Code and Codex CLI across SWE-bench Verified, SWE-bench Pro, SkillsBench, and TerminalBench, while consuming fewer tokens in most configurations. The harness supports 20+ frontier models across GPT, Claude, Gemini, and MAI families, plus open-source and local models, letting developers pick the right model for each task's cost and capability needs without being locked into a single vendor's tool. Lower token use translates to reduced API costs for equivalent work.

   GitHub's multi-model architecture enables cross-model critique (e.g., Rubber Duck, where one model reviews another's output), a capability single-model vendor harnesses cannot offer. The benchmarks show GPT models deliver the best value with strong resolution at lowest cost, while Claude Opus reaches the highest resolution at higher cost.