Multi-agent Systems Weekly AI News

March 24 - April 1, 2025

This week saw major advancements in multi-agent AI systems across industries. Microsoft unveiled new cybersecurity agents to fight phishing and data leaks, part of its expanding Security Copilot platform. Cisco launched its Webex AI Agent for 24/7 customer service, using human-like interactions to solve problems.

In logistics, companies like Amazon are using robot swarms that work together like a hive mind to move goods faster. BMW’s smart factories now have AI teams where different agents handle supplies, schedules, and quality checks. Jotform introduced no-code AI agents to automate forms for small businesses, making it easier to handle tasks like patient check-ins.

Ethical debates heated up as Klarna’s plan to replace workers with AI agent teams faced criticism. Researchers in Paris created an AI that predicts heart attacks 3 hours early by reading heart signals. Python developers got new tools like ChatAgent-py for building custom AI agents.

Security experts warned about risks when AI agents communicate secretly, which could lead to mistakes or hacking. Companies like Shelf.io shared blueprints for enterprise-scale agent systems that act like digital workforces. Redpanda raised $100 million to improve data systems for real-time AI agent teamwork.

Extended Coverage
Put an agent to work

Stop reading agent demos. Give one a job you repeat every week.

Describe the work, test the first result, and keep the agent available without running your own server.

Runs without your laptopBrowser + messaging appsBackups and clonesMemory survives restarts

Plans start at $29/month. Cancel anytime.

Hosted agent

OpenClaw or Hermes

saved state
Browser
WhatsApp
Telegram
Slack
“I checked the inbox, handled the routine messages, and sent you the one question that needs a decision.”
Create an AI worker that keeps running after this tab closes.
Open Agent Factory