#AI智能体
376 related articles
kosa8: The Saveable, Forkable, and Rollback-Ready Sandbox Built for AI Agents
kosa8 (COSA 8) gives each AI agent an isolated VM with its own kernel, supporting save/fork/rollback, human-in-the-loop approval, hash-chain auditing, and air-gapped deployment.
MCP Protocol Explained: How AI Connects to Any Tool or Data Source
MCP (Model Context Protocol) is the open standard that lets AI connect to GitHub, databases, and APIs. Learn how it works and why it matters for AI agents.
Strands Decider 2B: A Deep Dive into a Lightweight Open-Source Decision Model
Strands Decider 2B is a lightweight, open-source decision model built for agentic systems. This article explores its 2B-parameter design, open-source value, and potential role in Agent architectures.
AI Agents That Actually Make Money: Real Use Cases Enterprises Are Willing to Pay For
Which AI agent use cases are enterprises actually paying for? We analyze customer support, data automation, and content generation scenarios with measurable ROI.
Microsoft IQ Explained: Giving Enterprise AI Agents Real Business Context
A deep dive into Microsoft IQ — how WebIQ, Foundry IQ, Fabric IQ, and Work IQ give enterprise AI agents real business context, with Supply Autopilot demo and Agent 365 governance.
OpenAI Partners with Ironclad: How AI Agents Are Tackling Contract Automation
OpenAI partners with legal tech firm Ironclad to train and evaluate AI agents on complex contract workflows, advancing Computer Use capabilities in professional enterprise settings.
The Five Levels of a Software Factory: A Progressive Path to AI-Automated Programming
Architect Johnson's five-level AI software factory framework: from context engineering and loops to queues, orchestration, and the dark factory — and why true full autonomy doesn't yet exist.
Grokbot in Action: How 11 AI Agents Take Over a Bank Branch Manager's Paperwork
Grokbot claims to deploy 11 AI agents for bank branch managers in five minutes. We break down the multi-agent architecture, MCP integration, and real-world compliance concerns.
AI Agents Autonomously Launch and Trade Tokens: A Deep Dive into MCP, A2A, and the x402 Protocol
How do AI agents autonomously issue and trade tokens? A breakdown of on-chain practice using MCP, A2A, and x402: two agents with independent wallets complete a full token lifecycle on Arc chain — no human signatures required.
GrokBot Hands-On: Deploy 11 Backend Dev AI Agents in 5 Minutes
GrokBot claims to deploy 11 AI agents for backend devs in 5 minutes. We break down its setup flow, multi-agent coordination, and MCP protocol — plus an honest look at its limits.
Agentic AI and MCP Explained: From Generative AI to Autonomous Agents
A college workshop breaks down Agentic AI, Generative AI, and MCP (Model Context Protocol), with a live n8n RAG chatbot demo and tool picks like Cursor AI and Claude Code.
Hindsight: The Open-Source Tool That Gives AI Agents Long-Term Memory
Hindsight is an open-source AI memory tool with 25K+ GitHub stars. It gives coding agents long-term memory via Retain/Recall/Reflect and 4-way parallel retrieval, with Claude Code and Cursor support.
rea: Reverse Engineering Any Software with AI Agents — From App Behavior to Native Binaries
morluto/rea uses AI agents to reverse engineer anything from app behavior to native binaries. Gained nearly 3,000 GitHub stars in a day. Here's why it matters.
NVIDIA Open-Sources OpenShell: A Secure, Private Runtime for AI Agents
NVIDIA open-sources OpenShell, a Rust-based secure and private runtime for autonomous AI agents, gaining nearly 6,000 GitHub stars in one week. A deep dive into its design and significance.
Forge: An Open-Source, Self-Hosted Visual Builder for AI Agents
Forge is a MIT-licensed, self-hosted open-source visual AI agent builder. Assemble agents, tools, and RAG on a canvas with built-in observability, JMESPath token trimming, and flexible deployment.
dots Unveiled: Decoding the Always-On Agent Powered by GPT-6 Astra
A tweet claims the new AI product dots runs on GPT-6 Astra as an always-on agent. We break down what this means and why skepticism is warranted.
AI Agents for Small Businesses: Transforming Customer Acquisition, Product Building, and Financial Management
A new report reveals how small businesses are using AI agents for customer acquisition, product building, and financial management — plus hands-on training via ASBDC.
How to Choose Your Personal AI Agent: A Deep Dive into Four Key Decision Dimensions
Choosing between Muse, DOTS, Grokbot, Hermes, and other personal AI agents? This guide breaks down 4 key decision dimensions to help you pick the right one.
A Single AI Agent Burns 1.6 Billion Tokens Per Day: $100 Subscription Delivering $540K in API Value
A Reddit user reveals his single OpenAI Dot agent consumes 1.6B tokens/day, producing full CAD engineering packages — $540K in estimated API value inside a $100/month subscription.
Context Language Models: Letting AI Agents Manage Context Like Editing a File
Meta & UW propose Context Language Model, letting AI agents edit their context like a file instead of compressing it — preserving detail at the cost of prefix caching.