#33. Agentic AI in Education: My First Hands-On Test with Simular



Over the past week or so, I’ve been testing a new tool in the ever-evolving world of AI — Simular*. It's a browser-based platform that promises to bring “agentic AI” to life. The idea is compelling: instead of prompting a chatbot like ChatGPT to give you information, you can direct an autonomous “agent” to visit websites, scroll, click, read, extract data, and even write content like a newsletter or report. As an educator, this concept has obvious appeal.

When I started to test out Simular I had high hopes, especially in helping me streamline tasks like researching new EdTech tools, curating weekly digests for other educators, and summarizing long-form blog posts or research articles. In theory, it could serve as an automated research assistant — browsing the web while I focus on planning lessons for my students or doing other, related tasks. But after hands-on testing, I’ve come away with two key conclusions:

1. It’s not quite ready — at least for educators.

Despite its sleek interface and exciting potential, Simular is still more “smart browser” than true agent. Many of the key workflows I wanted to automate (like visiting multiple EdTech sites, collecting tool descriptions, summarizing them, and pushing the result to Google Docs) were stopped cold by login screens, CAPTCHAs, or limitations in Simular’s agent capabilities. The tool simply can’t bypass modern web security, which makes fully autonomous workflows unrealistic for now — even for tech-savvy users.

2. The future is bright — just not here yet.

That said, the concept of agentic AI has legs. I can absolutely see a future where a tool like Simular (or something like it) could autonomously build content collections, monitor EdTech trends, or even assist with lesson planning and marking. But that future is likely 6–12 months, or longer, away for mainstream educators. Right now, I guess it’s better suited to early adopters and developers who want to test the edges of AI. I guess my hands-on testing of Simular left me a little bit in what Gartner describes the 'Trough of Disillusionment' in the hype cycle of AI, whereby after the initial wild and overly-optimistic expectations of a new piece of tech comes a sense of frustration when using the tech.

In the video here I try out a few tasks with Simular. 




So what’s still working for me?

The good news is that ChatGPT continues to be the most reliable AI tool in my day-to-day practice. Whether, for example, I’m generating reading comprehension tasks for my Grade 4 students, writing video scripts for student broadcasts, or exploring how AI tools can be mapped to Bloom’s or SAMR models, GPT does the job — instantly, clearly, and with minimal friction. So, for now, I’ll stick with ChatGPT as my main AI assistant — and I’ll revisit agentic platforms like Simular later in the year. As these tools mature, I’m confident they’ll become more usable for educators. Until then, I’m staying grounded, curious, and focused on what works right now.

_____________________

*Simular is only available on Mac at the moment.

Comments

Popular posts from this blog

#30. Writing Tools: Three Simple Ways Educators Can Use Apple Intelligence in Notes

#29. Harnessing Text-to-Speech Technology in the EFL/ESL Classroom

#31. Reinventing Reading: Using AI and Digital Tools to Create New Formats of Educational Storytelling