Why AI products still need to be ‘designed’ for the people using them
Every week another AI product launches that promises to change how I work. Most of the ones I’ve actually tried have been pretty good at the clever part but pretty hopeless at the visual part. Logos that have been squashed, images stretched, and not a single master slide in sight! It’s a great starting point but it still needs a whole lot of ‘a designer’s eye’ to finesse. And if different teams are all using different platforms to create documents there’s no visual design system to make everything feel cohesive and on-brand.
Buyers and users are scanning for the same things we all scan for when we land on something new. Does this feel considered? Has anyone actually thought about my experience here? Most people won't tell you the answer though. They'll just stop opening the tab.
As the visual brand lead for Scail I’ve been building a set of Brand Guidelines for us to implement into all our work. We have been moving at such a fast pace, building documents and tools at lightning speed which is awesome in so many wonderful ways, but it highlights what often happens when there’s no design system or guidance. If we can give our AI tools these design guidelines from the start, then we can start to guide it into producing better quality outputs that feel much more cohesive. I still think a human eye should still oversee any outputs and designers still need to offer their input if brands and products want to feel cohesive and trusted.
Design works best when nobody notices
I'm not a SaaS product engineer, but I have spent decades thinking about how people experience the things they use, whether that's the packaging on a shelf, a printed brochure, a website or an app. The whole point of good design is to reduce the friction between a person and what they're trying to do. They reach for something, it works the way they expected, and they carry on with their day. Almost nobody notices when this is happening, which is exactly the point! The hard work has already been done well before they ever turned up.
AI product design has the same job, just with a much higher trust hurdle to clear. People aren't only asking "does this work?" They're also asking "is this safe?", "what is it doing in the background?" and "what has it crawled to find that image reference?" I've asked all three myself a fair few times this year! A clean visual interface alone won't answer any of those. You have to design for the worry, not around it. That means showing the AI's behaviour in plain words, putting controls where someone would naturally look for them, and being honest about what the system can and can't do yet.
Trust gets earned in the small moments
From what I can see in general consumer usability, the AI products that feel most usable right now are the ones that make their workings visible. They tell you the thought process, offer a quick way to undo, and if you ask it to, what data they used. They don’t ask you to read a ‘help’ article to feel safe. User trust in AI systems shouldn’t be something you bolt on with a reassuring badge in the footer. It’s something that gets earned along the way, when someone experiences a considered output and gives them an obvious way back if they change their mind.
That’s why human-centred design has to come before wider integration, not after. Pushing an AI product into more hands while those friction points are still there is the fastest way to lose the audience you were trying to win. From everything I've experienced in branding and B2B over the years, the businesses that pull ahead are usually the ones treating design, usability and adoption as one thing they're solving together, not three separate jobs being done in different rooms.
A clearer picture of where adoption is stalling
Most leaders probably don’t have a clear picture of where their AI products are losing users to friction, hesitation or mistrust. That is exactly what Scail’s AI Risk & Value Scorecard is built to answer. It’s a structured diagnostic that benchmarks a business across eight core areas of AI capability, including Integration and Adoption, Execution and Delivery, and Culture and Capability, three of the areas where AI product design and user trust in AI systems sit directly.
The Scorecard turns a fuzzy sense of “people will get used to it” into a clear picture of where adoption is actually being held back, where credibility is leaking, and what a 100-day path to a stronger position looks like. AI doesn’t earn wider adoption just because it’s available. It earns it the same way every product does. By being thoughtful about the person on the other end of the screen.
Read more about our AI Risk & Value Scorecard.
