WSJ What’s News AI Agents Like OpenClaw Are Here. How Can You Use Them?
73 snips
Mar 29, 2026 
Isabel Bousquette, a WSJ tech reporter on AI and Silicon Valley trends. She explains what agentic AI can do today, from coding help to customer service. She dives into OpenClaw and full-access agents, their productivity promise, security and liability risks, and how companies weigh costs versus benefits.
AI Snips
Chapters
Transcript
Episode notes
Agents Execute Tasks Not Just Answer Questions
- AI agents act in the world by executing tasks on a user's behalf, moving beyond just answering queries.
- Isabel Bousquette contrasts coding agents used by engineers with customer-service agents handling order status and simple account actions.
Token Costs Are Weighed Against Massive Productivity Gains
- Companies measure agent value by token costs versus productivity, sometimes finding token spend exceeds salary.
- Firms report engineers using agents can be 10x more productive, so higher token costs may still be justified.
OpenClaw Enables Full Assistant Access With Big Security Tradeoffs
- OpenClaw is an open-source orchestration framework that aims to turn agents into broad personal assistants with access to many systems.
- Isabel warns full access is a huge security risk because agents can delete files, emails, or expose credit-card data.