AI is breaking free — an emerging AI-UX trend

Note — to be honest, this has been in my drafts for 1.5 months now. Holiday season got to me? Also, back when I published the Trending AI-UX Patterns” ebook, I wrote about this specific pattern. It was a small signal then, but now it’s gaining traction. Not mainstream yet but definitely picking up momentum.

Quick refresher —

We started with a conversational UX, the first user experience allowing us mere mortals to the experience the power of AI. As the technology became more standardised, more use cases started coming up. Product teams then started using it with the traditional UI layouts, as it happens when a new material/technology arrives. 

How arrival of technology combines with UI, ref. from the “Trending AI-UX Patterns” ebook

For example, Elicit’s “parallel thinking” AI-UX is an example of how AI can be used with the old UI paradigm. A beautiful implementation. Even V7 Labs’ Go product uses an Excel type interface to run AI operations parallely. Clay AI is another enterprise example with a similar AI-UX.

I like to refer to this parallel thinking pattern as “living documents” — 

Living documents pattern page, ref. from the “Trending AI-UX Patterns” ebook

While humans only do a few things at a time, AI can do 100s. Running things parallelly is a pattern that’s so powerful, it feels underutilised. Main use case has been extraction of data, but across multiple attributes — defined by the user.
~ “Living Documents”, Pattern 07 of Trending AI-UX Patterns.

However, this article is about a new trend, a trend that’s breaking free and going beyond the limits of traditional UI.

Parallel thinking in a non-traditional UI

Product teams have started to realize — to use AI to its full potential, it need not be used linearly, like we humans are used to when doing tasks. There doesn’t have to be a step 1, step 2, … etc. You can create a linear process, hand it off to AI and then run multiple linear processes parallelly.

The real power of AI is bulk, but we’re used to linear.

And that’s where the traditional UI fails. Well, sort of. 

Traditional page-based UI doesn’t work when you’re trying to multi-task. Or when you’re trying to look over or control parallel processes. The experience should be able to provide the flexibility of getting a bird’s eye view, hopping between processes quickly, and diving deep into one of them for further analysis, with the ability of zooming out again.

Anything come to mind? Infinite canvas. 

Infinite canvas pattern page, ref. from the “Trending AI-UX Patterns” ebook

What if you could play with AI in an infinite canvas? Outside the bounds of GUI, this AI-UX pattern uses AI accelerate the user in a 2D space. The user’s intention when using this pattern would be escape the limits set by devices and replicate the physical environment tasks digitally.
~ “Infinite canvas”, Pattern 08 of Trending AI-UX Patterns.

Pattern: Infinite canvas

Replicating the real world in a digital setting; being able to group and combine pieces spatially. 

How are products moving towards infinite canvas?

1. Conversational apps

While ChatGPT, Claude and other conversational UX are powerful and used by millions, it doesn’t always capture how humans think; non-linearly. Sometimes, I just want to start a new thread mid-conversation or view different possibilities from one point and then select the direction of the conversation. Current chat-interface are limited. How does an infinite canvas help?

Note — All hyperlinks below contain a demo video of that product. I got you.

Subform’s non-linear exploration of conversation is aligned to how we connect dots and ideas. *that linking-string animation Oooo*

Screenshot of Subform’s UX / Source: Aiverse

Isn’t research also non-linear? Heuristica’s trying to achieve just that by providing a lot of options (and i mean a lot!) to research in ALL possible directions.

Screenshot of Heuristica’s UX / Source: Heuristica demo

2. Creative tools

This one might be an obvious application. Creativity for sure cannot be contained in grid, boxes and buttons. So shouldn’t our tools allow us to “explore”, as we do in real world? Create, combine, and be spatially creative.

From using the power of AI in a discord channel (initial days of Midjourney) to today, it’s been a wild upgrade.

Visual Electric allowing creatives to explore without any bounds. Multiple functions to apply to an image, and an infinite canvas to not limit the possibilities.

Screenshot of the Visual Electric’s UX / Source: Aiverse

Concept by Adobe is the closest a product has come to that captured the true essence of remixing. It’s literally as easy as playing with clay, drag & drop an image as style or composition and voila.

Interaction of dropping a photo in Concept / Before / Source: Aiverse

Interaction of dropping a photo in Concept / After / Source: Aiverse

Ideoform is another great example. You can combine images by creating an overlaying layer and merging them together (with a prompt if required).

Ideoform’s remixing images / Before / Source: Aiverse

Ideoform’s remixing images / After / Source: Aiverse

3. Enterprise products

You know you’re lacking behind if big MNC’s have also started implementing the user experience right? While the exact concept of infinite canvas hasn’t made into their product-lines, their implementation is quite close.

What do enterprises like? 
Dashboards. 
How do you implement AI x Dashboards x Infinite Canvas? 
Widgets.

Flexible enough to show any data in any visual, 
fixed enough to show them in a bento grid.

Meter’s generative UI hybrid approach is actually very interesting. Using a familiar conversational style for the user to create anything with AI, and then allowing them to drag & drop the output as widgets on an infinite canvas. Genius?

Meter’s conversational (left side), infinite canvas (right side) and drag/drop functionality / Source: Aiverse

Salesforce’s latest lighting canvas is as creative as an enterprise may get? I absolutely loved the implementation — really powerful and flexibile.

Salesforce bento dashboard / Source: Salesforce blogs

Salesforce bento dashboard / loading state / Source: Salesforce blogs

4. Software creation

I believe this to be the next space where infinite canvas will make its mark. Imagine being able to create software, as easily as playing LEGO, in a digital space. 

Devin’s quad-grid based approach gave us a glimpse of what the future of software building would look like; controlling multiple processes parallelly and getting a overview at all times. 

Devin’s AI software engineer quad-grid UX / Source — Linkedin

However, there’s still another step to be taken. Tldraw Computer is a venture in that direction. Being able to create software like diagrams, and visually knowing what the flow of a function is or where something breaks.

Creating apps in Tldraw computer like diagrams / Source: Tldraw Computer

And that’s the complete deep dive into pattern 08: Infinite Canvas.

Want more AI-UX patterns?

Give the Trending AI-UX Patterns ebook a read. It has ALL patterns, emerging and mainstream, extracted from the best companies designing for AI. 

Any one of these patterns could be the game-changing feature for your product. After all, “chat” was also a UX pattern that made ChatGPT viral.