Build an AI Form-Filler That Works on Any Website

Filling out repetitive forms is one of the most soul-crushing tasks on the web. Job applications that ask for your work history after you’ve already uploaded your resume. Support tickets that demand the same account details every single time. Lead forms, onboarding questionnaires, registration flows: the same fields, the same answers, over and over. An AI form filler running in the browser can handle this for you, with an understanding of the form that traditional autofill does not have.

The problem with repetitive forms

A knowledge worker fills out many web forms every week. Each one demands a context switch: find the right information, type or paste it in, reformat it to match the field’s expectations, and submit.

It adds up. Here is what a modest habit of forms can cost in a week:

Beyond the time, manual form-filling introduces errors. A mistyped email address, a wrong date format, a skipped required field: small mistakes with sometimes significant consequences. Automation solves both problems at once.

Why autofill hits a ceiling

Browser autofill and password managers are the most common attempts to solve this problem, and they’re useful as far as they go. But they hit a ceiling fast.

• Browser autofill only handles a fixed set of fields (name, address, credit card). It can’t adapt to custom field labels, multi-step forms, or fields that need a written answer.
• Password managers (1Password, LastPass) extend autofill slightly but are designed for login credentials, not general-purpose form completion.
• RPA tools like UiPath or Automation Anywhere can fill forms programmatically, but they need professional setup and dedicated infrastructure, which is overkill for individual or small-team use. If you’re not a developer, browser automation for non-developers is a more practical starting point.
• Copy-paste libraries save snippets of text but still need manual selection and placement: they organise your clipboard, they don’t fill anything.

None of these tools understand the context of a form. They can’t read a label like “Briefly describe your relevant experience” and write a tailored answer. An AI form filler can:

How it works: the page plus a language model

Agentic Workflow combines two capabilities to make AI form-filling possible: access to the page and LLM reasoning. The browser extension can read every element on a web page, including form labels, input types, placeholder text and field names, even on websites that don’t have an API. It passes that structure to a local or cloud model, which decides what value belongs in each field. The values are then typed into the page, as if you had typed each answer.

Because everything runs in the browser, form data does not leave your machine unless you add a node that sends it somewhere. With a local model (WebLLM in the browser, or Ollama on your computer), even the AI step stays on your device.

Build your AI form-filler, step by step

The quickest start is the docs example Fill Form from JSON. It keeps your values in one node, asks before filling, types each value into the matching field and submits the form. No AI yet: it is the reliable base you add AI to.

Add the AI step

To make the filler understand any form, put an AI step between your data and Form Fill:

1. Add Get Forms Schema after the trigger to read the fields of the form on the page.
2. Add a Basic LLM Chain with a model (a local Web LLM or a cloud model) and a Structured Output Parser. In the prompt, give it the list of fields, your profile, and ask for a JSON object of field id → value.
3. Switch Form Fill to Manual mode and map its values to the model’s output.

This is where the AI reasoning happens: the model works out that “Full Name” is your name, that “LinkedIn URL” is your LinkedIn, and that “Why do you want to work here?” needs a written answer based on the company description it can read on the same page.

Test and refine

Run the workflow on a test form first. Check the filled values and adjust your prompt or profile data as needed. Pay attention to fields with strict formats (dates, phone numbers) and to optional fields you may want to skip. Once it performs reliably, save it and run it with one click whenever you meet that form.

Advanced: AI that understands the whole page

The real power of an AI form filler shows when the model sees the full page, not just the form. For a job application, it can read the job description on the same page and tailor every answer to it. For a support ticket, it can read your previous ticket from a connected data node and keep the case details consistent. A static autofill tool cannot do this.

Where it helps

• Job applications: fill application forms, cover-letter fields, screening questions and availability with answers tailored to each role
• Lead generation forms: fill contact and company information quickly and consistently across many platforms. For outreach at scale, pair this with a strategy for automating LinkedIn prospecting without getting banned
• Support tickets: pre-fill issue descriptions, account details and troubleshooting steps from your own data
• Vendor registrations: onboard to supplier or partner portals without re-entering the same company information every time
• Survey and feedback forms: automate routine internal surveys where your answers are predictable

Try it

Browser autofill was built for a simpler web. Today’s forms are complex, context-dependent and often multi-step. A workflow that reads the form, picks the right value from your data and fills every field in seconds, with you approving first, turns that data entry into one click.