A few weeks ago, I had a chat with a friend and facilitator (Arthur from WorkshopWednesday) about the rapid progress in AI technology. As avid Miro users, we couldn't help but wonder, "Will Miro jump on the AI train and introduce some new features?" It didn't take long before our question was answered - the very next day, Miro announced the beta of Miro AI. Curious to see its potential, I eagerly tested it out to answer the question: Gimmick or game changer? Let's take a look.
How can you access Miro AI?
Miro AI is currently in beta and you can sign up here for the waitlist.
Once you have access, you might find yourself slightly confused about how to access the AI. In comparison to other AI tools, there is no single place to access the AI. Instead, the AI is spread out throughout the Miro user interface. But this makes sense because there are multiple content elements that an AI could use as input.
There are currently two main ways to interact with Miro AI:
- Through the command palette (Shortcut CMD + K)
- Through the context menu of supported board elements
Each supported board element will display a shiny blue button, that is impossible to miss. Miro AI currently supports the following board elements:
- Sticky Notes
- Mind Maps
- Code blocks
Having experience with other AI tools such as GPT-4, Midjourney, and Stable Diffusion I’ve took a closer look at how Miro implemented AI for each of these board elements.
What Miro AI can do today
Add similar sticky notes
Sticky notes are an essential building block in Miro, acting as the primary content-generation tool on the board. Although Miro introduced an AI clustering feature last year, it hasn’t quite taken off. (At least that’s my impression). With Miro AI, however, we get some new capabilities - one of which is generating similar sticky notes.
How it works: Select a single sticky note, and press the big blue button. Wait for the magic to happen and boom! You got some new similar sticky notes.
I think the intention behind this feature is to enrich brainstorming sessions. However, I found the results to be okayish. After some testing I came to the conclusion that the AI only considers the content within the sticky note you selected. In the example below, you can for example see that i've prompted the AI to give me similar sticky notes to "Movie night". While technically the results are good, not all of them are relevant for the context of an "office party".
At first, you might wonder: Why is that a problem? In my experience, the more detailed and context-rich your prompt is to a GPT AI system, the better the results. So when you only send the AI a tiny sticky note, it of course generates something, but often it doesn’t fit to what you had on your mind.
I would have assumed that Miro would pass additional context or data from surrounding board elements into the prompt. But that doesn’t seem to be the case for now – but I’m sure they are already thinking about it.
Summarizing sticky notes
This feature immediately got my intention, because that’s a task that I do quite often and that requires quite some manual work. But I do have to say, that I bumped against the same problems as with generating similar sticky notes.
In the example below, you can see how I tried to summarize feedback for Anna. Since the feedback is only in the three sticky notes, I selected only those and prompted the AI to summarize. Interestingly, it mistakenly attributed the feedback to Max. Max was one of the participants on the board, so maybe Miro does send along some data?
But still, AI doesn’t get what I want to achieve here and clearly there is a big “Feedback for Anna” sticky note on top. I’ve tried to include that sticky note, but it didn’t improve the results.
What I would love to do, is to give additional context into the AI prompt such as “I want to summarize the feedback for Anna”, or “Summarize the most common user feedback in these sticky notes”.
Generate an image
This feature lets you select a sticky note, and Miro AI will generate an image based on the text of that sticky note. This won’t knock your socks off if you are familiar with AI tools like DALLE 2 and Midjourney. But it’s neat to have these features right inside Miro, because Miro is a visual canvas that allows one to quickly organize and iterate on multiple different image generations.
One problem I see here is that sticky notes are not made for adding a lot of details. Their purpose is to capture ideas and thoughts in a concise format. But if you want to generate good images, you need to add quite some details and be very specific about what you want to achieve. So these two don’t really go together naturally.
None the less, it’s a great feature to have inside Miro! And since it comes bundled with Miro, it might save you the OpenAI subscription.
Remove background image
This is a game-changer feature for anyone who creates presentations in Miro. Select an image, press the magic button, and Miro AI will remove the background flawlessly (well, most of the time). It couldn’t be easier.
I’ve spent hours doing this in Canva and then exporting the files to Miro. So I’m delighted that I can skip all of that and directly remove image backgrounds where I’m building my slides.
Expand Mind Map
Miro AI's Mind Map feature is perhaps its most intriguing aspect. When expanding a Mind Map, the AI provides a visually appealing map that elevates the output, making it more coherent and organized.
The only drawback I encountered is that the AI only considers the context of a single node in the Mind Map, which can lead to lower-quality results. It would be advantageous if the AI considered the entire Mind Map as input, which I hope will be addressed in the future.
Generate a code block
Miro has a code block with syntax highlighting, and although I haven't found much use for it, others may find it more valuable. By incorporating AI-generated code snippets, users might find it helpful to discuss and document code directly on the board. However, users who need this feature will likely use their programming environment (like Microsoft Copilot).
Cards in Miro include elements for tasks, user stories, and organization in kanban boards. Miro AI allows you to generate user stories based on existing card phrases, which could be useful for specific project management and user research contexts.
My verdict: Gimmick or game changer?
In my opinion, the current version of the product (which is also in beta) is more a gimmick than a game-changer. While it's a starting point, I think it lacks the ability to customize prompts and a more intelligent way of adding context to improve the results. Without these two features, I don't see myself using it much.
But I must say, I see a lot of potential with Miro. In my experience, one of the most challenging aspects of working with AI is being overwhelmed with information. For instance, if you use ChatGPT, all the responses will be in a linear flow. After just a short time, you may become lost in the jungle of words provided by the AI.
In comparison, Miro provides an infinite visual canvas that is inherently non-linear. Everything you create with Miro's AI is easy to move, organize, and connect with other content on your board. This makes it a promising medium for interacting with AI in an efficient and less frustrating manner.