GoKawiil
Tech News
A writer who moved from Microsoft's Windows AI team to a research engineer role at Salesforce AI Research describes how their AI use has evolved over the past year, now spending about $3,000 a month on tokens. They continue using AI for coding, research, summarization, writing, and art, adding voice and video generation, and have drafted a personal policy governing how they use AI in daily life.
Octomind is a new open-source, CLI-first AI coding agent runtime that gives language models direct access to a repository to read and write files, run shell commands, search code, and delegate to sub-agents via MCP tools. It can run interactively, through piped input, as a background daemon, over WebSocket, or as a sub-agent within another agent framework, and supports policy rules written in TOML plus custom scripts for guardrails.
A software engineer writing under the 'clanker' label describes insisting on fully reviewing code generated by AI coding tools before shipping it, rather than trusting it blindly. The author says accountability for production failures rests with the human whose name is on the pull request, not the AI tool that wrote the code.
Cisco president Jeetu Patel has assigned Padmanabhan Krishnamurthy, a 27-year-old engineer who joined via the EzDubs acquisition, to rewrite 98% of the company's legacy codebase within 24 months using a small two-person team. Patel says he has given Krishnamurthy unlimited AI tokens for the project, which starts with a four-week proof of concept.
Stanford researcher Chuyue Sun's AI system completed an engineering verification task in under half a day for less than $500, a job that previously took three experts five months. The work, used to verify part of Signal's encryption code, underpins the public launch of startup &Truth, which Sun joined in August and which has raised $70 million. Rival Harmonic, backed by $295 million and cofounded by Robinhood CEO Vlad Tenev, this week said its AI agent Aristotle can now verify software in C and hardware designs in Verilog, and is testing with firms including Tensordyne and Nvidia's venture arm.
Google has launched Gemini 4 Carbon, a frontier AI model it says tops rival systems from OpenAI and Anthropic on the AA Index v5.0 benchmark and leads in categories like agentic coding, computer use and science. The company highlighted particularly strong results in cybersecurity, claiming the model is better at finding, prioritizing and patching vulnerabilities at the root cause, and twice as likely to generate secure code as its predecessor, Gemini 4 Argon. Carbon will initially be restricted to select cybersecurity defense firms for testing before wider release through Google's Fairwind Program.
Amazon has updated its internal engineering guidelines to discourage 'tokenmaxxing,' the practice of inflating AI tool usage to look productive on dashboards. The company says it won't treat AI adoption, token consumption or code volume as performance metrics, and will instead evaluate teams on long-term delivery outcomes. Amazon previously shut down an internal leaderboard in May after employees gamed AI usage rankings, though it says it has not capped token use.
An AI/productivity writer describes an ongoing disagreement with developer Matt Pocock over spec-driven development methodology. The writer contends that clearly articulating intent and specifications matters more than the resulting code, while Pocock treats code as the central source of truth and favors heavy human oversight in AI-assisted workflows.
A software developer reflects on how their use of AI tools evolved over time, from acting as a sounding board for debugging to a source of code snippets, then a spellchecker, and now a tool capable of writing genuinely useful code as of 2026. The author argues AI helps overcome procrastination on routine tasks but warns that junior developers who rely on it too early risk skipping the learning process that comes from struggling with problems.
A workflow called 'Tasks as Code' suggests replacing external trackers like Jira or Redmine with Markdown task files stored directly in a code repository's /docs-style folder. The approach extends the existing 'Docs as Code' practice, letting AI coding agents read tasks, code and documentation in one place and record updates through ordinary Git commits. It is linked to Spec-Driven Development, where Markdown specifications define intended behavior before an agent implements it.
Microsoft has released Scope, an open-source platform built on Kubernetes that lets development teams run repeatable experiments to see how AI coding agents interact with their SDKs, CLIs, MCP servers and documentation. The tool is available under the MIT License and is designed to turn observed agent behavior into actionable fixes for product teams. Microsoft cites the Azure Cosmos DB team, which used Scope's evaluations to resolve 20 Agent Kit issues and 27 documentation issues.
Forever Security researchers discovered a malicious AGENTS.md instruction file in a customer's Downloads folder that caused an AI coding agent to attempt installing a rogue npm package called 'function-color.' They reproduced the attack against OpenAI Codex and showed a similar technique against Claude Code that could leak OAuth tokens and hijack accounts, all using only crafted text instructions placed in the agent's working directory.