@sayoni_dutta_roy Congratulations on the launch! 🎉 I love that you’ve focused on empowering non-tech users, it is an important space to address. You’ve done an awesome job creating a user-friendly platform, but I think there’s still some room for improvement to make it even more intuitive and accessible.
Editing mode.
Using block names as node names might be a huge simplification for users. This will reduce visual complexity by 50% because you could remove the white part of any block and leave only the grey one. It also will help users who aren't familiar with concepts of unique identifiers and variables. From a tech perspective, you will need to update agents' instructions in case of block renaming and the same names will be prohibited, but it looks like a small price for perceived simplicity.
If you also show all (X) buttons only on hover, schemes will look outstandingly simple.
A few potential adjustments to the top panel:
- An Undo button (with a shortcut key) will be extremely useful
- The "N nodes to fix" warning could focus users on a particular error block
- Maybe rename Run into Test, after all, it is its purpose
Running mode.
It was challenging to understand the difference between Run and Bulk Run because Run still could work with data arrays. WDYT about moving the Import CSV button from the second tab to the first one? After pressing the button, you can block the input field using different methods.
This way users won't need to think about a bunch of logical entities: Bulk Run, Add Run, Run All.
This will also help with the table at the bottom of Bulk Run. It took me some time to understand that inputs there work the same as inputs from the Run tab.
I think it'll be useful to completely hide output until the results are ready. Outputs look exactly like inputs and might create the impression that a user has to write something there. The download button is also not necessary without data.
Terms.
My suggestions about used terms are quite intuitive and might be completely wrong.
Node -> Segment or Block or Module ("node" sounds scary and unclear)
Description -> Personality (otherwise, you need to think of how to distinguish this field from the Task field)
Expected Output -> Expecter Results (this is less technical)
Pipelines + AI Agents -> AI Agent + Subagent (I've got an impression that Agent became a common term for a finalized AI product, so it's weird to see this term as one of the building blocks)
Overall, I think you’re off to a fantastic start. I'm sure GenFuse AI will become even more approachable for non-tech users. Amazing work!
@anton_osipov Wow. This is AMAZING feedback. Thanks Anton. I really appreciate it. We will definitely make these changes soon.
Again, thanks a lot :)
Report
Congrats on the launch of GenFuse AI, @sayoni_dutta_roy! 🎉 It looks like such a cool tool that makes creating AI automations super easy. Love that it's made for anyone, no technical skills needed!
Congrats on the launch. How customizable are the workflows? For example, can users integrate external systems or tweak the behavior of individual agents to better fit their specific needs?
@vishal_bansal003 We take security very seriously. All data is encrypted in transit and storage. We will also support on-premise deployment soon. So the data never leaves your servers
TRDATA
GenFuse AI
GenFuse AI
GenFuse AI
ArticleBunny
GenFuse AI
GenFuse AI