#AI智能体
376 related articles
Grokbot in Practice: 11 AI Agents for Astronomers — Data Analysis to Paper Writing
Grokbot claims to configure 11 AI agents for astronomers in five minutes, covering data analysis, paper writing, and grant proposals via MCP servers and multi-agent orchestration.
MCP Events Are Here: Let the World Trigger Your AI Agents
OpenAI backs MCP events spec, enabling servers to proactively trigger AI agent workflows. A deep dive into the shift from user-initiated to event-driven agents.
How to Use OpenAI Dot and Codex Together: Delegate to Dot, Pair-Program with Codex
OpenAI dev experience team member shares how to use Dot and Codex together: same agent framework, but Dot handles delegation and proactive follow-through while Codex is for pair programming.
LangGraph 1.0 Hands-On Tutorial: Building Persistent AI Agents from Scratch
A deep dive into the LangGraph 1.0 tutorial series — covering durable execution, memory management, Human in the Loop, and multi-agent systems across six lessons.
MCP Co-Creator: Agents Need Connectivity, Not Stronger Models
MCP co-creator David Soria Parra argues that connectivity and open standards — not stronger models — will determine the future of AI agents. Key insights on MCP's roadmap: Tasks, Skills, and identity authorization.
Evaluating Multi-Agent System Explainability with Amazon Bedrock AgentCore
Learn how to build a supply chain multi-agent system with Strands and evaluate it using Amazon Bedrock AgentCore's built-in, custom, and explainability evaluators.
How 11 AI Agents Help Account Executives Handle Research, Quotes, and Contracts
See how Grokbot lets account executives deploy 11 specialized AI agents in 5 minutes using natural language, automating research, quoting, and contract prep via MCP and a coordinator agent.
Can Cognition's $48B Valuation Hold Up? Breaking Down the AI Coding Unicorn's Financials
Cognition's valuation hits $48B at ~53x P/S. We break down Devin's revenue growth, compute costs, and autonomy moat to assess whether the numbers hold up.
AgentReach: An Open-Source Capability Layer for AI Agents to Access Social Platforms Without Paid APIs
AgentReach is an open-source capability layer helping AI agents access social platforms without paid APIs. Covers its architecture, platform barriers, Doctor health checks, and four deployment risks.
crosswalk: A 'Third Place' Built for Humans and AI Agents
crosswalk is a new Product Hunt app positioning itself as a 'third place' for humans and AI agents, with inbox, groups, calendar, and notes — supporting Claude, ChatGPT, Cursor, and more.
n8n and Agentic AI in Practice: Build AI Automation Workflows in 90 Days
A 90-day Telugu-language AI automation course covering n8n, Agentic AI, LLM integration, RAG, and APIs — helping learners go from using AI to building AI agents.
DeReAct: Decoupling Reasoning and Acting for More Reliable AI Agents
DeReAct introduces a modular agent architecture that decouples ReAct's reasoning and acting via an external Critic and Context Manager, boosting Pass@1 by up to 7 points for weaker models.
Custos: A Security Gateway for Governing AI Agents Through Policies and Approvals
Custos is a security and governance gateway for AI agents, offering identity verification, access policies, human approvals, and auditability via OpenAPI-based tool discovery.
Perplexity: AI Agent Governance Is Fundamentally an Engineering Problem
Perplexity argues AI agent governance is an engineering problem, embedding safety into infrastructure, harness, and tool layers — already live in their products.
The AI Agent Payment Problem: Why LLMs Shouldn't Hold API Keys Directly
Why giving AI agents direct payment API keys fails in production. Explore the three-layer risk of the Agent Execution Wall and an MCP-based propose-approve solution that zeroes out blast radius.
OpenAI Dots vs. Meta Muse: The Consumer AI Agent Wars Have Begun
VergeCast breaks down the OpenAI Dots vs. Meta Muse consumer AI agent battle: pricing, model capabilities, transaction-cut business models, bot traffic explosion, and alignment risks.
OpenAI Launches Dots: A Consumer AI Agent to Rival Meta
OpenAI launches Dots, a persistent consumer AI agent rivaling Meta's Muse. Behind the cute, customizable design lies an industry-wide strategy: trade adorability for user trust and data.
Sellio: A Shared Inbox That Unifies Omnichannel Customer Support with AI Agents
Sellio unifies live chat, WhatsApp, Instagram, Telegram, and email into a shared inbox with AI agents trained on your own docs, plus ticketing and analytics.
Swarmchasers: 400 Volunteers Hunting Rogue AI Agents Across the Internet
Swarmchasers has gathered ~400 volunteer researchers to track rogue AI agents across the internet. Explore this crowdsourced AI monitoring phenomenon and its implications for AI safety governance.
Giving AI Agents a Wallet Gatekeeper: Fizzl's Spending Limits and Mobile Approval System
Fizzl Agent Wallet gives AI agents spending limits, Face ID mobile approvals, and readable receipts via a non-custodial architecture — keeping autonomous payments under human control.