For a long time, Automatio has lived as a visual no-code tool. We built it around a Chrome extension and a drag-and-drop builder so people could scrape data and automate the web without writing code. But recently, we’ve been working on the next logical step: an AI-driven automation layer.
The goal is to move from "building" a workflow to simply "describing" it.
Testing on Real-World Use Cases
When it came time to test the new AI agent, we didn't look for abstract problems. We went straight to what our users have been doing for years: automating online polls and forms. A lot of people use Automatio for things like school competitions, community challenges, or team polls. It’s a very common use case, so it made sense to use these as our playground to see how the AI handles dynamic UIs and session control.
Historically, you’d have to manually select the elements and map out the clicks in the builder. With the new AI upgrade, you just talk to it. You can say: “Go to this poll and vote for option three fifty times,” and the agent identifies the elements and handles the loop on its own.
If you want to see the specific breakdowns of how this looks in practice, I wrote these guides for the most requested platforms: StrawPoll and Poll.fm
What’s Next
Polls were just the first use case to prove the concept. The next phase is much broader. We’re moving toward a future where any repetitive web task, whether it is re-doing massive online forms or complex data extraction, can be handled by simply chatting with the agent.
