Intention is all you need. Copilot × Figma to democratise business processes?

As a UX designer and ex-developer, I’ve often dreamed of a future where my sophisticated Figma prototypes could be transformed into functioning products, dodging lengthy full-scale development processes. There have been many attempts at rapid application development (RAD), but they've had questionable success. These systems captured instructions, sequences of steps, UI elements, but lacked an understanding of the designer's intentions. There's a huge gap between the "what and why" of a designer/BA, and the developer's "how exactly".

Then I saw "Record with Copilot" at Microsoft Sydney, and it absolutely blew my mind. For the first time, a software design system could understand the user’s intention beyond just a sequence of recorded actions. I feel that, sooner rather than later, there will be a solid AI bridge over the costly, innovation-stopping gap between the "why" and the "how".

Design by storytelling

When designing a solution, I usually plot a story around some customer's problem, create hi-fi prototypes in Figma with very life-like examples of data on screens, and then record a narrated screencast video. I explain the problem, outline the user's expectations, and showcase how to solve this problem, step-by-step, highlighting the value of each idea.

That's how product owners, customers and other business-side stakeholders get an intuitive feel about the solution on top of a clear understanding. It also gives the developers all the necessary details on how exactly it should work and look, and an opportunity to raise technical concerns early.

Whether the video focuses on just a small feature, or a novel UI approach, or a full-blown end-to-end business process, the very visual, almost tangible presentation makes people excited and feel that the solution is really possible, and the innovation is really achievable. Pretty efficient, right?

"Record with Copilot" works in a very similar way. Essentially, you record a screencast of yourself performing tasks with existing software (like copying and pasting from Excel, filling out forms, or attaching documents to emails). It’s similar to the old good MS Office Macro Recorder or macOS Automator, but with a one significant difference: while recording your actions, you

explain with your plain voice what you're doing and why, what you're trying to achieve

– just like I narrate my videos. And Copilot gets it.

Like my human stakeholders, AI is able to understand that the recorded sequence of actions is just an example, an illustration of a meaningful business process that has an intention, an objective, and variabilities. Therefore, it can generalise and build a solution that would work in real life – rather than just a rigid script.

Democratisation of business processes

One of very prominent successes in business democratisation was Microsoft Teams. It gave people who were already working together an easy, bottom-up way to self-organise into workgroups and securely share documents – without the need for updating the organisation’s tree, involving sysadmins to create shared folders, or relying on managers as gatekeepers to control access (which rarely works smoothly).

With Copilot, this same bottom-up approach is being applied to automation.

Sure, some might argue that you can’t create anything truly sophisticated with what seems like a "macros recorder on steroids." Fair enough! But at every presentation on business automation, the standard pain points are the same: siloed, disjointed data spread across numerous systems, improper tracking of customer interactions, endless copy-pasting between poorly integrated systems, and so on. And then, lots of people, bored to death thus making costly mistakes, are needed just to keep things functioning daily.

Now, with Copilot, you can easily build a smart AI agent to automate these simple tasks – quickly and without much investment. And with minimal involvement of the developers who always are overloaded and who wouldn't be able to attend to your needs because you are too low on the highly competitive priorities list.

Instead of going through the lengthy and often bureaucratic software development process, one can automate yourself or with minimal help from BAs and UX designers. This is especially valuable for simpler solutions where products already exist and don’t require extensive development. You just need to: 

• Identify work that's already being done somehow (should it really be done?)
• Uncover the logic of the work and simplify it to a decision tree (most manual processes have simple logic, right?)
• Make it all explainable (you'd have to explain it to a BA and the devs anyway, right?)

And then the magic happens. Even if your manual process has low priority amongst other business processes and "hot" topics. Even if it is off the radar. Even if "we haven't resources for that till next year". Even if it was planned but then got excluded because your project stuck in MVP state forever.

Designer's dream

But what if we could take this a step further? There are many tools beyond Copilot that help write code and accelerate developers' work, but I’m talking about something very different – a way for designers to see their ideas accelerated. This isn’t about going low-code; it's not about creating lifelike Axure prototypes that a developer would need to reconstruct from scratch. It’s about designers designing and coders coding, each doing what they do best, but without important details lost in translation just because what's important for a designer isn't equally important for a developer, and vice versa.

As a UX designer, I already create narrated videos of my Figma prototypes to share with stakeholders and developers. These videos showcase problems, user expectations, and proposed solutions, helping everyone understand and get excited about the vision. But what if I could share them with Copilot?

What if Copilot could take my Figma designs and narrated videos and then actually build a working prototype directly from them? Not being limited by existing software's UI that "Record with Copilot" is currently restricted to. Not constrained by the primitive logic of simple tasks.

Don't get me wrong, this isn’t about eliminating developers; it’s about enhancing the process. Developers wouldn’t need to build everything from scratch – they could start with a quite functional prototype that already captures the design intentions, that already looks pixel-perfect, and just needs further implementation. Copilot could be the tool that finally bridges the gap between a designer’s vision and a working product. It could break the vicious cycle where important details get excluded from MVPs, leading to the failure of innovation.

While we’re not there yet, the foundation is being laid. This is the first time we’ve seen a product development system begin to understand intentions beyond just instructions, and that’s a huge leap forward. The potential here is enormous, and I’m excited to see how much closer we can get to turning product designers' dreams into reality.