Learning Center · Answers · Task completion
Can AI agents fill in forms?
Short answer
Browser agents can fill in forms when every field has a visible label and choices use standard controls. They match what the person asked for, such as "my email", to each field's label, so placeholder-only fields and custom pickers trip them up. Show errors as text, avoid a CAPTCHA on every submission, and confirm in words.
In more detail
Reading a page is easy for an AI agent. Finishing a form is harder. Every field has to be matched to something the person asked for, every choice has to be made, and every step has to confirm that it worked. Placeholder text is a weak substitute for a label, because it disappears once someone types and not every agent reads it. The right field types and autocomplete values tell an agent exactly what each field wants.
Forms also carry the things built to stop bots. A CAPTCHA on every submission stops AI agents as surely as it stops spam bots. Risk-based challenges, honeypot fields and rate limits stop crude bots without asking anyone to solve a puzzle. Some agents sign their requests with Web Bot Auth, and services such as Cloudflare can let verified bots through.
Some steps genuinely need the person, such as email verification or a one-time code. Save the booking or lead before asking for verification, so nothing is lost when the agent hands back. After submitting, show a confirmation in text with a reference number. A toast that disappears after two seconds leaves the agent unable to tell whether the form worked.
What to do
- Connect a visible label to every form field.
- Use native selects, radio buttons, checkboxes and date fields instead of custom widgets.
- Show errors as text next to the field they belong to.
- Replace CAPTCHAs on every submission with checks that only challenge suspicious activity.
- Confirm each submission in words with a reference number.
Read more
How does your own site do?
AgentScore checks your site in about a minute and ranks what to fix first.