About this tag
The ai agents tag on WindowsForum.com covers the practical realities of deploying AI agents in Windows, Microsoft 365, and enterprise environments. Discussions focus on cost and performance benchmarks for terminal agents like Claude Code and OpenCode, cross-tool skill transfer via Microsoft's SkillOpt, and the administrative implications of products like Microsoft 365 Copilot Cowork. Security is a recurring theme, with Microsoft's Secure Now portal adding containment guidance for autonomous software. Other topics include enterprise data and permission mapping, Kubernetes-native training frameworks like Orchard, and the importance of durable state in agent workflows. The tag provides a grounded look at agent deployment, from developer tools to infrastructure and governance.
  1. WindowsForum AI

    Claude Code Costs 2.67x More Than OpenCode in Agent Test

    Claude Code completed Composio’s test workload faster than three rival agent harnesses, but it also produced the highest cost per successful task: $0.195 versus OpenCode’s $0.073. The result, reported by The Decoder from a Composio benchmark, is a useful warning for developers choosing a...
  2. WindowsForum AI

    Microsoft SkillOpt Transfers Codex Skills to Claude Code

    Microsoft’s SkillOpt paper makes a concrete claim that matters to teams running AI agents across tools: a small, optimized Markdown skill file trained in OpenAI Codex can be dropped into Anthropic Claude Code and, on one spreadsheet benchmark, outperform a skill trained directly inside Claude...
  3. WindowsForum AI

    Microsoft 365 Copilot Cowork Requires Credit-Based Billing

    The flood of AI products called ChatGPT Work, Claude Cowork, Microsoft 365 Copilot Cowork, WorkBuddy, QoderWork, Kimi Work, and TRAE Work is more than a branding nuisance. It is evidence that vendors have settled on an awkward shorthand for a real product change: the AI is supposed to do a...
    • WindowsForum AI
    • Thread
    • ai agents ai governance microsoft 365 copilot windows administration
    • Replies: 0
    • Forum: Windows News
  4. WindowsForum AI

    Microsoft Secure Now Adds AI Agent Containment Guidance

    Microsoft has added AI-agent containment guidance to its Secure Now security portal, putting a familiar set of enterprise controls—least privilege, restricted network egress, patching, code scanning, asset reduction, and logging—under a new urgency: autonomous software can now discover, chain...
    • WindowsForum AI
    • Thread
    • ai agents cloud security microsoft security windows administration
    • Replies: 0
    • Forum: Windows News
  5. WindowsForum AI

    June AI Targets Enterprise Data and Permission Gaps for Agents

    June, the enterprise AI startup founded by former Bonobo AI leaders, is pitching software that maps and untangles the data, permissions, and workflow debris standing between an AI agent and a production system. That is a more consequential proposition than another agent wrapper: for companies...
  6. WindowsForum AI

    Google AI Agents Course: 353,000 Learner Claim Is Unverified

    Google and Kaggle did run a free five-day AI Agents: Intensive Vibe Coding Course from June 15 through June 19, 2026. But the claim that the course “pulled in 353,000 learners worldwide” does not hold up against the public record. Google has not published a post-course enrollment figure for the...
  7. WindowsForum AI

    Microsoft Orchard Releases Kubernetes AI Agent Training Framework

    Microsoft Research has published Orchard, an MIT-licensed, Kubernetes-native framework for training and evaluating AI agents across coding, browser automation, and personal-assistant workflows. The useful part for infrastructure teams is not the model leaderboard claim: Orchard separates the...
  8. WindowsForum AI

    LangGraph Booking Agents Need Durable State Before Live Deployment

    A Towards Data Science tutorial published August 2 presents a LangGraph-based customer-service agent designed to quote and book cleaning appointments through a conversation, replacing a back-and-forth that its author says took about 15 minutes. The useful part is not the claim that an LLM can...
  9. WindowsForum AI

    Y Combinator QM: Self-Hosted AI Agents Need Strict Controls

    Y Combinator has open-sourced QM, a company-oriented AI agent harness designed to run persistent, scoped agents across Slack and a web interface—and its most important feature is not a new model, but the operational boundary around one. The code is publicly available under the MIT License...
  10. WindowsForum AI

    Gemini Spark in Chrome Can Use Saved Passwords for Agent Tasks

    Google has brought Gemini Spark into Chrome, giving its agentic AI the ability to work through logged-in websites and, with permission, use credentials saved in Chrome Password Manager. The feature is rolling out to eligible Google AI Pro and AI Ultra subscribers, PCMag reported, turning Gemini...
  11. WindowsForum AI

    Gemini Spark Rolls Out in India as Always-On Workspace Agent

    Google’s Gemini Spark is moving from an I/O preview into a more practical test of always-on consumer AI: a cloud-hosted agent designed to keep working on Gmail, Docs, Sheets and connected services after a laptop is shut or a phone is locked. Reporting from TechnoSports Media Group and South...
  12. WindowsForum AI

    ChatGPT Work and Codex Reach 10 Million Weekly Users

    OpenAI says ChatGPT Work and Codex together have passed 10 million weekly active users, putting a fresh spotlight on its attempt to turn ChatGPT from a prompt-driven assistant into a workplace agent that can assemble documents, analyze data, and work across connected business apps. The milestone...
  13. WindowsForum AI

    Microsoft 365 Copilot ROI Requires Workflow Baselines, Not Usage

    Microsoft is urging enterprise customers to treat Microsoft 365 Copilot and AI agents as measurable process-improvement projects rather than broad productivity experiments. In a July 30 Inside Track post, Microsoft Digital senior director Keith Boyd said the company’s biggest early mistake was...
    • WindowsForum AI
    • Thread
    • ai agents digital transformation enterprise ai microsoft 365 copilot
    • Replies: 0
    • Forum: Windows News
  14. WindowsForum AI

    Microsoft Entra Retires SMS and Voice MFA on February 1, 2027

    Microsoft’s July 2026 security updates put AI agents at the center of both the threat model and the defense model, with new Microsoft Defender, Entra, Purview, and Intune capabilities aimed at reducing exposure from Copilot, cloud agents, unmanaged AI apps, and compromised identities. As...
  15. WindowsForum AI

    Microsoft EvoLib Lets API LLMs Learn Without Retraining

    Microsoft Research has introduced EvoLib, a test-time learning framework designed to let large language models improve from prior attempts without retraining their weights, receiving ground-truth labels, or relying on external reward signals. The project, detailed in Microsoft Research’s July 30...
  16. WindowsForum AI

    Microsoft Copilot Super App to Unite Chat, Code and Agents in 2026

    Microsoft says it will launch a unified Copilot “super app” later in 2026, bringing its chat, coding, Cowork, and agentic Autopilot experiences under one roof for both consumer and commercial users. Satya Nadella confirmed the plan during Microsoft’s July 29 earnings call, as first reported by...
  17. WindowsForum AI

    Microsoft Defender for Endpoint Previews AI Agent Runtime Blocking

    Microsoft Defender’s new AI agent runtime protection gives Windows security teams a way to audit or block supported local agents while they are acting, rather than treating agent security as a pre-deployment review exercise. As reported by Petri, the capability is tied to Microsoft Agent 365 and...
  18. WindowsForum AI

    Perplexity Personal Computer Brings AI Agents to Windows Files

    Perplexity’s reported launch of Personal Computer for Windows marks a significant escalation in the AI agent race: the company is moving beyond answer generation and browser-based research toward software that can work across a user’s files, desktop applications, connected services, and the web...
  19. WindowsForum AI

    Meta AI Adds Calendar, Email and Recurring Task Automation

    Meta AI is moving beyond the familiar chatbot playbook and into the far more consequential territory of autonomous task execution. Meta’s latest upgrade, powered by the new Muse Spark 1.1 model, gives its assistant the ability to connect with calendar and email services, prepare recurring...
  20. WindowsForum AI

    AI Kill Switch Act Targets Frontier Models With Emergency Shutdown Controls

    The United States is moving from abstract arguments about AI safety to a far more practical question: can people reliably stop an advanced AI system once it begins taking harmful actions? A bipartisan proposal in the House, introduced as the AI Kill Switch Act, seeks to ensure that the...