Building a Platform That Scales With You

AI can accelerate your workflow, but it takes a platform that is trustworthy, intuitive, and accessible

Will Dunlop
Resultid Blog
6 min readNov 14, 2022

--

The rapid implementation of AI technology has changed the way we work forever. So much so that the level of automation AI brings is no longer just a benefit, it’s necessary to make your workflow more efficient. The possibilities for how we can accelerate the way we work are endless with AI technology, which is what makes thinking about the future of work so damn exciting. At Resultid, we’re particularly interested in leveraging AI technology to streamline the process of analyzing large amounts of messy, qualitative data. But we don’t just want our users to simply use our product — we want our users to take the flexible tools we offer and combine and customize them to best suit their unique needs. With all the power and potential that artificial intelligence (AI) systems can bring, we have a responsibility in providing such tools to design them to ensure that they create the best possible outcomes for our users.

Designing AI-Powered Tools That are Trustworthy, Interpretable, and Accessible

One of the core challenges faced with using AI in this process-integral way is that it can sometimes be difficult to understand and explain the AI systems' results. If something goes wrong, it can be difficult to chalk it up to anything in particular. For us, a huge no-no is an outcome where a user is left scratching their head when something goes wrong. 🚫 But mistakes do happen, and when it does a transparent and understandable user flow can improve the user experience. Therefore, we aim to build trust in our system by connecting the user with what is happening at each step.

We believe that the most valuable AI-powered feature we can provide isn’t a “black box”, or a system that takes from a user and spits out some (hopefully useful) output. Instead of having a one-sided workflow, we want our tool to be a conversation, with an understandable back-and-forth that can ensure the best and most repeatable results. The user should understand how different components, arrangements, and settings can influence the way that our tools produce results, instead of mindlessly accepting output from a rigid system. This way, users are at the heart of everything we do and are included at every step. 🫶

With Resultid, the user can customize their experience by keeping control over aspects of the model that are meaningful to their result so that they can adjust the types of results they get. Think of our features as lego pieces. Sure, each set comes with specific instructions, but pieces can be reused to customize Lego works of art to your heart’s content. 🤩 Our dedication to the user’s ability to be tightly integrated translates to a more useful output for our users who don’t want to attribute their conclusions to something “Resultid found”, but “what they found using Resultid.” ✅

Designing AI-powered Tools That Are Intuitive

Take a look at our stunning Theme Discovery narrative, which extracts the most prevalent themes from qualitative datasets. This tool is great for quickly summarizing large datasets, but making this narrative accessible means presenting an explainable interface that delivers intuitive identification of groupings of qualitative data items to a user. On top of that, the narrative had to be designed to help users understand the process involved, how they can adjust our tool to best suit their needs, and what these results actually say about their data. People aren’t mindreaders, and this is not as immediately intuitive! The mapping of this capability onto user behaviors is way more complex, with users caring about different definitions and combinations of groups.

Thankfully, our small but mighty team loves a good challenge. 💪 To achieve this, we gave users control over the number of themes identified, as well as insight into what fundamentally defines a theme. This way, users don’t have to think of our themes as some new format that supersedes or replaces their existing processes; instead, they can as closely as possible replicate and accelerate their processes with our tool.

Consider a use case where a user researcher at a SaaS company is interested in identifying specific insights about a platform from qualitative feedback they’ve collected. This user might try out different configurations of our clustering tool to arrive at something that fits closest to their notion of an established “insight” — which may be as specific as a single search bar often bugging out, or as general as users noticing that the platform runs slow. Whatever the result may be, you’re looking for a conclusion that will drive your business forward and for us it’s about being part of your journey — not the actual destination.

We’ve conducted extensive research on our user base, and have found that the workers who look for specific software platform issues are using the same tools as workers looking for the big-picture trends in social listening data. We want to provide users the ability to apply AI models and tools to their work in the ways that fit best with their existing workflow, whether or not they have clean, organized data, or technical knowledge about the AI systems. Compared to traditional platform services, which might provide a set of tools that confine users to a specific process and platform, we break down the barriers to getting data in and out.

Fundamentally, we never want to replace any user’s workflow. Instead, we want to accelerate it. Our tool is built around transparent, easily understandable building blocks that build user intuitions, not try to replace them with blind faith in some black box of AI. What this means in practice is that users don’t have to put complex models and data preprocessing steps together, they can take their familiarity with their own workflows and use our tools to make their workflows scale better for them.

As the future of AI and automation drives full speed ahead, we hope that learning about your qualitative data practices will help our tools improve. More than that, we’d love to hear about it! 👂If you’re interested in sharing your practices with us or learning more about how your needs can shape our product, join our fantastic Council or shoot an email over to bertie@resultid.com!

--

--