Exploring the Potential of Designing With AI

Applying Design Thinking to AI

The adoption of AI products came swiftly to nCino. There was an enormous push to develop new AI experiences in all of our product lines. My ask was to work on an experience which would change the way users uploaded documents.

At nCino, documents would be uploaded to the document manager one at a time and manually sorted into the correct placeholder. The process was tedious and there was a significant amount of customer feedback to imply that we needed to find a way to speed it up.

Enter Banking Advisor, nCino’s native AI agent. My task was to help shape an experience that utilized the convenience of AI to sort uploaded documents in bulk.

Click on each image to view larger

  • AI functionality has the potential to revolutionize banking software. Providing a high level, summarized overview of pages of information would be an enormous value add.

    Locate & File would allow users to offload a lot of the grunt work involved with bulk uploading documents and sorting them into the right locations.

  • Innovations in automation don’t come without some drawbacks. While designing the Locate & File experience I considered the lack of control with the AI agent and whether the feature would actually speed up the process or cause unnecessary confusion.

    I took careful consideration in how I designed the product to try and account for as many potential issues in the user’s experience as possible. Through this project I learned a lot about about using direct, descriptive labeling, providing a linear path through the workflow, and offering manual override options.

Workflows & Wireframes

Wireframing an AI experience can quickly become a very circular pursuit. AI creates a lot of potential new branches for any experience, and each of them needs to be considered and weighed against the purpose and value of the product.

An AI product that can do everything isn’t particularly useful for any single task, but an AI tool with direct parameters and source material can change an entire industry.

Locate & File Workflow

Challenges

Hallucinations
We needed to account for any mistakes Banking Advisor might make. While powerful, it wasn’t always 100% accurate.

Information Hierarchy
Deciding which actions were the most important to include and in what order they belonged proved to be a bit of a challenge. It made us reconsider the very framework of our document filing system.

Low On Physical Space
As an offshoot of the document manager, which was built in a different framework and couldn’t be updated, we needed to contain the experience inside a modal

Ease of Use
Though many people have adopted the use of AI in recent years, a lot of bankers still maintained technologically simple workflows and had little experience with the tool.

Final Experience

No matter how much we iterated on the product, it always felt like a work in progress. I believe this is due to a lot of the unanswered questions we had while developing the product. A lot of AI is still new territory for many industries, and it certainly felt like we were testing the waters.

Qualitative research would be the necessary next step to determine how customers were responding the the product, the workflow, and what parts of the experience were helping them.

Determining the placement of actions and descriptive labels in the information hierarchy proved to be a challenge. How could we efficiently and seamlessly guide the user through such a complex process?

We needed to make sure there were manual options for the user to engage with. The experience includes a manual search and assign.