Arro is an AI-powered UX research platform. Customers are UXR or Product teams across different scales of companies that try to understand their users with the quality of a 1:1 interview and the quantity of a research form. Thanks to AI, they can offer personalized interviews to users to grasp their complete feedback while doing it at scale in a matter of days. Once the feedback is received, AI also assists them in summarising the learnings, extracting opportunities and generating next steps.
I started working with the Arro team in February 2023 when Craig Watson reached out. Before that, Craig had sold his startup to Spotify and then joined the music company as a PM for a few years. I joined them part-time as a Founding Designer and got the opportunity to design the brand, the website, the product, and motion design.
To learn more about Arro, visit arro.co.
1. The Brand
When I joined the team, the first task was to design the brand. We knew the product was about UXR, but AI was not the main driver at the time. With experience, I knew that founders tend to be quite opinionated about the look and feel they want to give to their brand. So, I sent three drastically different suggestions of visual identity to grasp the vibe.
These three suggestions covered very different approaches. But there were all approaches I loved because they helped the founders to find their vision for the brand. Ultimately, they settled on going for the first option, with a few tweaks—the main one being to work on the logo to add an arrow (of course).
After a few explorations and tweaks, we settled on a final visual identity.
2. The product
In the meantime, I was working on the product as well. The first step was to set up a design system to ensure we could move quickly. I started working with custom designs, but it was taking a lot of time to do the product + the design system at once (some unused designs are shown at the end of the article.) So, we decided to use the Untitled UI library and customize it. Since everything was moving simultaneously, there was not much to customize except for some colors and font styles.
The first version of Arro did not have any particular AI feature. Before we switched to an AI-powered product, I started designing the core of the product (settings, studies, and study page creation).
The following screens were designed more than a year ago, so there might be a few issues with some exports or contents. We were also using Atlassian as an imaginary customer, but the company has no relationship in any way with Arro.
I liked this design. It was nothing special, but we achieved designing something that looked good in a very short amount of time. Of course, we had no time to build all of these features, but the pace was terrific.
[AI enters the chat]
Things got more complicated then. AI was everywhere, and a startup not building AI features was meant to die. We decided to shift to an AI-powered UXR platform. The newest AI capacities were perfect for the UXR use case. The interview was not a classic interview anymore but a written AI interview, so I had to change a few things. Half of the effort was now spent of making the Human Interaction with AI delightful.
The interview was now a conversation. The interview summary was now a text-based summary with insights and actionable items. And ultimately, since the UX for typing a long answer is somewhat bad (even if you have the most delightful UI in the world), the answers could now be given with voice.
3. The website
After a few months of building, the product was robust enough for us to start onboarding users. So Craig asked if I could set up a website - in under a week. I used Webflow and got started the same day. There's nothing more exciting than knowing the website will be live in a couple of days, whatever its form, especially as a perfectionist myself. It pushes you to make it work as well as possible.
There was very little research. No A/B test. No landing pages for Ads. We basically needed a showcase on the Internets. And it lives on arro.co.
As usual, I spent way too much time on the hero section. But I love to. It's basically our chat UI, but for users to sign up for the waitlist.
The rest of the website's work consisted mostly of visual graphics that did not necessarily match the UI perfectly, but hey, it was done in four days.
You see a bento grid of features at the end of the page. I have nothing against bento grids, but I usually find them hard to read through. So, I had the idea to opacify out the other cards when hovering over one of them. This little trick made its way quietly to @ridd_design Twitter at some point without me knowing.
4. The motion design
For the launch of Previews, Craig asked me to create a short video for our social media. Previews let you preview your conversation blocks before publishing them. Before that, users had to publish the conversation and test the live version to experience the flow. This feature allowed them to go through the interview without creating false data.
Again, I had a few days to create a video, so I decided to use the UI and some of my 3D skills on After Effects.
I was quite pleased with the final motion design. The interaction design was smooth, and the feature was clearly explained. Although adding a voiceover would have enhanced the video further, time constraints did not permit it.
Conclusion
Designing a brand and product from scratch as a sole designer is the opposite of what you learn at school. There's no time for A/B tests or a perfect SEO strategy. You have to follow your intuition and build something that feels like the right thing. With this in mind, I think we've successfully built a brand and a product that look good and have a great potential for Product and UXR teams.
I can't wait to see where this will lead the team, and I can only thank Craig for this extremely valuable experience.
-
PS: some extra designs that we haven't used
