That AI tools radically changing the productivity of digital product designers feels obvious, but what is perhaps less obvious is that that the largest change will be procedural rather than creative. Once the hype dies down around generative AI creating content and we move beyond the showmanship, AI-driven tools that focus on productivity will emerge. These will run in the background and just get stuff done. They’ll enable new ways for us to interact with computers and digital services.

A prediction:

In the near future AI driven Sketch-to-UI design tools will radically simplify and speed up the product design process. They will do this by being able to interpret a designer’s sketches and turn them into production-ready interfaces using pre-designed and pre-built components in (AI-ready) Design Systems.

For organisations able to take advantage, this will reduce the product design and build phase from days and weeks to minutes. New design and research practices will emerge to take advantage of the new opportunities and design governance will become even more critical.

The trend in managing complexity in the design process

• Tesler’s law “states that for any system there is a certain amount of complexity which cannot be reduced” . Complexity is preserved.
• A hundred and ten years ago Ford created their famous production line. Complexity was moved from individual workers to the production line itself and defined processes. Building a car is no less complex (Tesler’s Law) but each worker does simpler jobs but the system of the production line is more complex. Ford invested in solving the hard problems once and the result was more cars being produced.
• Design Systems have done the same for designing digital products. Creating a Design System – a set of pre-designed, pre-developed, pre-tested components – is investing in complexity up front.
• They manage complexity in the design process, store it, and remove it from the day-to-day tasks of production designers. Production designers do simpler tasks, have narrower scope. The result is more design gets produced.
• It’s reasonable to expect this trend to continue and more efficiencies in the design process to be created. We should expect more of the complexity of product design to be moved up-front to systems, processes, defaults, and frameworks.
• AI tools are likely to enable a large acceleration of this trend. Product(ion) design will become much simpler while Design Systems will become correspondingly more complex.
• (And finally: beardy old men will continue to moan about the good old days when design was design)

Two demos of AI-driven sketch-to-UI design tools

AirBNB

This demo from the AirBnB design team – showing a computer recognising the intent of very simple sketches then assembling a user interface from the design system – deserves to be up there with the great HCI demos from the likes of Xerox Parc.

This is from 2017 – why are we still working like it’s 2016? 2017 was a long time ago in the digital product design profession, Figma was just emerging and design systems in their mature form were just starting to appear. AI has got cheaper and more widely available and design systems are pervasive. The conditions are right for picking up these ideas and running with them.

https://airbnb.design/sketching-interfaces/

TLDraw MakeReal

6 years later, a demo of a similar process got attention on the socials. Ultimately this one feels less ambitious but demonstrates that people are interested in exploring the opportunities created by emerging tech in this space. And – powered by ChatGPT – it shows how off-the-shelf LLMs make hacking in this space cheaper and easier.

Sketching to deployable in real-time

These demos are cool – and point to a near future where sketch-to-code is a real thing. Getting the AI to write the code is cool. But getting it to assemble components from a design system will be the key to using AI to generate production-ready interfaces. If the AI tools can only use carefully crafted, tested and already-deployed components to build the UI, the risks and quality concerns associated with AI code approach zero.

An AI-enabled sketch-to-code design process might look something like this:

1. 🧑🏽 Roughly sketch a user journey (On a whiteboard in a meeting with stakeholders?)
2. 🤖 AI design tool auto-generates the interfaces by assembling components of the design system.
3. 🧑🏽 Use verbal prompts to suggest content
4. 🤖 AI design tool generates content guided by the style guide
5. 🧑🏽 Adjust the designs by adjusting the component choices created by DesignAI
6. 🤖 Deploy to live (or other testing/prototyping/research environment of choice)

We can move from days or weeks to minutes. We will be able to deploy to live before we’ve left the meeting*. This will be a wild step-change increase in designer productivity. This will both require and enable new design and research practices. It will put new importance on DesignOps, Design Governance, and the Design Systems that support this speed.

*We might not always want to deploy from a sketch, but the ability to do that opens up options

How will this change design?

That’s a title I can’t start to fully answer! But here’s a starter-for-ten list:

• Organisations that can harness AI-driven design tools will reap large benefits in efficiencies and speed-to-market. But this will have costs in continued investment in upstream processes and design governance.
• How design works alongside product and engineering will change. New boundaries and ways of working will be created. (I suspect that design may lose influence here given our broad lack of ability to talk business).
• Using these advanced systems will become an increasingly specialised design role. It will be hard to move between roles that use them and those that don’t. This is a continuation of a current trend.
• Design systems will become more complicated and necessarily inseparable from front-end engineering. The increased complexity will make them harder to change. Building design systems to be consumed by AI-driven sketch-to-UI tools will become a highly valuable skill.
• The tools we use will change. Adobe’s $20 billion for Figma will be a distant memory.
• How we think about prototyping and user testing will change. If we can have have fully working journeys available instantly how will that improve how we test designs? If we can change prototypes instantly in response to user feedback how does change our design processes? (If we can change live products instantly response to feedback what does that mean! 😱)

It’s gonna be fun/scary!

Is AI coming for our design jobs?

Yes. No. Both. AI enabled design tools have the potential to fundamentally change design processes. New design roles will be created, current ones will be disrupted. The role of designers in an AI-driven design future will be different. There is potential for this to be as big a shift in design ways-of-working as Macs and Photoshop were in 80s. (I don’t think it will be quite like that, but it’s plausible.)

• Yes – it’s coming for our jobs because design will be cheaper and quicker to create. So each ‘unit’ of design will take fewer designers.
• No – there’s not a fixed amount of design to be done in the world. If you make design cheaper it will stimulate demand and more designers will be needed to meet that demand. (This has been true of design in the last decade, as digital design has become more streamlined teams have got bigger not smaller).
• Both – at some point the efficiencies and demand will reach and equilibrium, if means more of fewer design jobs would be a wild guess. But my wild guess is more.
• Both – What I’m covering in this article is only one section of digital design – product(ion) design for largish companies with high design maturity. Design practices outside of that won’t be affected in the same way, and more product design being done may increase the demand for surrounding design jobs.