Is this AI agent legit? Nine checks before you pay
An AI agent is legit when you can name the company behind it, read how it handles your data, see a clear price and find reviews from people who used it. If any of those four are missing, slow down. The nine checks below take about thirty minutes and cover all four.
New AI agents launch every week, and many of them are made by small teams. Small is fine (some of the best tools I use come from two or three people). What you want to know is whether the team is real, whether the product does what the page says, and what happens to your data. These are the checks I run myself.
1. Find the company and the people
Scroll to the footer and look for a company name, an address and a way to contact a person. Then look the founders up. A real team usually has a history you can see on a professional network, a code repository or earlier products.
2. Read the privacy policy and the terms
You are looking for three answers. What data does the agent store, who else receives it (the model provider is usually one of them), and can you delete it? A privacy policy that answers those in plain language is a good sign.
3. Check what the agent can access
Agents ask for permissions: your inbox, calendar, files or browser. Read the permission screen before you click accept. Grant the smallest access that still lets you test the product, and use a test account when you can.
4. Find the pricing page
A legit product tells you what it costs, or tells you how to ask. Look for the billing period, what counts as usage, and how to cancel. If the only way to see a price is to give your card for a "free" trial, set a reminder to cancel before it renews.
5. Read reviews in more than one place
Look for reviews on independent review sites, community forums and directories, alongside the testimonials on the product's own site. Read the three-star reviews first. They tend to describe what the product does well and where it struggles, in the same paragraph.
Watch for patterns that suggest reviews were bought: many five-star reviews posted in the same week, all with similar wording, from accounts with no other activity.
6. Look for a public changelog or roadmap
A team that ships regularly usually writes about it. A changelog, a blog or release notes with dates tell you the product is maintained. Dates also show how fast the team fixes problems.
7. Try the product on a task you already know the answer to
Give the agent a task where you know the right result. Ask a research agent about your own company, or ask a writing agent to summarise a document you wrote. You will see quickly whether it is accurate, and whether it says so when it is unsure.
8. Ask how it handles mistakes
Send one question to support: "What happens when the agent does something wrong, and how do I undo it?" The speed and quality of the answer tell you a lot about the team. A good answer covers logs, undo or approval steps, and a contact person.
9. Start small and keep an exit
Pay monthly before you pay yearly. Export your data early so you know the export works. Keep the agent away from payments and customer messages until it has earned your trust on smaller tasks.
A checklist you can copy
- I can name the company and at least one person behind it.
- The privacy policy says what is stored, who receives it and how to delete it.
- The agent asks only for the access my test needs.
- I know the price, the billing period and how to cancel.
- I read reviews from at least two places other than the product's site.
- The product has dated release notes or a changelog.
- It got a task right that I already knew the answer to.
- Support answered my question about mistakes.
- I am on a monthly plan and have tested the data export.
Questions people ask
How can I tell if AI agent reviews are fake?
Look at timing, wording and the reviewer's history. Real reviews arrive over months, mention specific tasks and include some criticism. Bought reviews tend to arrive in bursts with similar phrasing.
Is it safe to give an AI agent access to my email?
It can be, if the company is real, the privacy policy is clear and you can revoke access at any time from your email provider's security settings. Start with a secondary account and move to your main inbox once you trust it.
What should I do if an AI agent charged me after the trial?
Contact support first with your account email and the charge date. If that fails, your card provider can help with a dispute. Next time, set a calendar reminder the day before any trial ends.