What can AI agents do (and not do)?
Updated
Short answer: An AI agent is a model that plans and takes actions with tools: it can research, write and send drafts, call APIs, run code, update spreadsheets and CRMs, and repeat tasks on a schedule. It works best on bounded, checkable jobs with a human approving anything risky.
| Task | Agent-ready today? | Example |
|---|---|---|
| Daily research digest / newsletter draft | Yes, with human send | setup |
| Crawl and score a catalog | Yes | setup |
| Publish pages from a data file | Yes, with review | setup |
| Pay for API calls | Emerging (x402), use spend caps | trend |
| Send money, sign contracts, message customers | Keep a human in the loop | - |
What can AI agents not do (yet)?
They are unreliable on long, open-ended work without checkpoints, they can be misled by untrusted content, and they should not move money, sign agreements, or message people on your behalf without approval.
What is the difference between an AI agent, a chatbot and an automation?
A chatbot answers. An automation (Zapier, n8n) runs fixed steps. An agent decides which steps and tools to use toward a goal.
What can AI agents do for my business?
Common starting points: drafting a newsletter, keeping a directory or listings up to date, triaging inboxes and support, research and reporting, and lead lists. See real setups at dartzdabz.com/setups.
Want tomorrow's movers by email?
One short email a day: new agent tools, the biggest trust-score moves, and one build you can copy. Free.