Year
2024
Client
IBM
Category
Product Duration
16 weeks
Create a system that can fit to an expansive generative tool that will have multiple usecases. Oh and make it pretty.
Overview
Our nimble team of two needed to take IBM’s Garage usecase and redesign the interface to be easier to understand and to be distill complicated actions that differ from flow to flow depending on the user’s need. Also, we needed to design an interface that could be flexible enough to accommodate radical differences between starting and end points that the user would choose between.
Essentially taking the prompting part of A.I. chat and making it more data oriented as well as refining it to Turbotax levels of easy step by step use.
Process
We created an initial dashboard. We needed a hub to allow for filtering and selection of all the different flows. Flows could be a product planning flow to a generative AI tool that would refine an idea for future product features. We also needed a way to specifically filter out A.I. enabled tools for companies that wouldn’t want to use this tool with them.
Each tool needed to have a description of all of the features, outcomes, and even example outputs that could be attained by using IBM Advantage.
After the dashboard, we took a look at the IBM digital product design process and assess the different stages. We took what we knew and researched user behaviours to establish a planning system that allowed us to stack multiple flows ontop of each other in a sequence.
We had to create safeguards that made sure certain flows that depended on one another notified the user if it was possible to proceed or if more data was required for upload.
The FLow
Once past the planning stage, the left hand navigation system helped track the user where they were in the current process as well as letting them jump to certain parts if able.
Data upload (seen above) needed to accomodate the uploading of multiple data sources that could be analyzed by A.I. Each piece of data was analyzed in succession to one another before it was added.
Skipping ahead to where the rubber meets the road, generated ideas would be displayed as cards and allowed the user to quickly browse and view the extended amount of information for each card.
Idea refinement would occur when the user chose the corresponding button and could chat with A.I. to get clarity on the result as well as see some more options/refinement of the selected card(s).
For the final overview after the user went through all steps, we constructed an overview and suggested follow up page. This used the A.I. to determine proper next steps in DPE and retained data for later use in the process if necessary.