AI/ML are powerful drivers of magical experiences. That is of course - when everything goes to plan and they get it right. When they don’t, - which is less often than when they do, but users are not forgiving -  it can create significant trust issues. It is key when working with intelligent systems to always always include the user in the conversation. At People.ai I worked on creating multiple touchpoints within the web platform to do just this.

​

​

MATCHING ECOSYSTEM

A key aspect of the core product was automatic activity capture. Each activity was  then 'matched' to the appropriate account or opportunity record in the app and the users CRM.

​

When things didn't line up from the users viewpoint the immediate inclination was to distrust the data. We needed to create an easy frictionless way for them to make edits as well as simultaneously gathering feedback that would help improve the accuracy of matches. The experience I created for this purpose resulted in helping users easily correct any issues themselves while also significantly lowering the amount of time internal teams spent assisting users with matching issues. I also worked on a supplementary experience that helped investigate and explain why an activity matched to a given record.

​

WHAT KIND OF MEETING IS THIS?

In another example, I learned through research that it was valuable for users to know how they were spending their time. More particularly what types of meetings were they spending their time on? This meant needing to have meetings classified by type eg: Demo Meetings, Sales Events, Regular Team Meetings etc. The only way to get this information currently was manual input which is tedious and often incomplete. I worked with the data science team to create a low barrier feedback collection UI within the product to learn and train the data model to improve our accuracy at automated meeting classification. Despite a lot of skepticism, the UI was highly engaged and proved users want to be a part of the conversation.

​