Enterprise apps, from CRM to ERP to marketing automation, have long had a common structure.
Whether in the cloud or on-premise they are anchored by a database layer, responsible for storing records concerning customers, products, and other entities. At the top is a presentation layer, responsible for displaying lists or summaries of the underlying records, and facilitating the entry of new records. Between these two layers is business logic that determines how to aggregate records, what restrictions to put on record entry, and the like.
As a user of these apps you predominantly use the app itself: You log in and manipulate the presentation layer directly to enter or see the information that’s relevant to the task at hand. You might go into a CRM system, navigate to your list of customers, sort by the last date that you contacted them, then reach out to those folks that bubble to the top. Or you log into your marketing automation app and start entering parameters to configure your next campaign. Apps with good user experience make it easy for you to see or enter information relevant to what you think you should be doing.
We’ve been trained to interact with apps in this way for decades. So all of the above is painfully obvious. But it’s also all about to change as A.I. permeates the world of enterprise software.
Fundamentally it’s a transition from presentation to conversation, and it will happen in three stages.
The first stage is most familiar: content generated or selected by machine learning models integrated into the classic apps. Now when you view a list of subscribers they have a color-coded badge indicating their likelihood to churn, and a top product to recommend to them. The traditional presentation layer embeds A.I.-driven calls to action.
In the second stage, A.I. identifies valuable tasks and pushes notifications to users suggesting they take action. For a recent example, a medical benefits management company had A.I. evaluate claims before any human intervention, approving those that were clearly appropriate. The tricky claims are pushed to nurses and doctors for further review. With apologies to Marshall McLuhan, at this stage the message is the medium, not the app’s presentation layer.
Whether it’s email or a rich messaging platform like Slack, the notification serves as a gateway deep into the presentation layer of the app that generated it. A churn alert links directly to the affected customer -- you no longer navigate down to that user from the app’s landing page. And recipients can even directly take action within the message itself as part of micro flow. The app is still the system of record, but its presentation layer is no longer the entry point for many user tasks.
In the third stage, the messaging capabilities of the app evolve via natural language processing to become interactive bots. “Show me a list of customers similar to this one.” “Is now a good time of year to offer a promotion to those customers? If so, schedule one for the best day next week.” Your conversation with the app is the focal point of your interactions. It’s two way: Sometimes the app notifies you about something that underlying models believe deserves your attention, and other times you ask for specific information or for the app to take certain actions.
At this point the transformation is complete. Natural language processing and machine learning are now core components of the binding between your conversation and the database.
So call me bullish on Slack -- rich messaging platforms that allow for micro flows stand to become the dominant UI of future enterprise applications. Hurdles remain -- as anyone who’s been down a modern phone tree is aware, bad natural language processing is infuriating; periodically I’d like to throttle Expensify’s concierge bot.
As these UX challenges are whittled away, however, the most fundamental shift in enterprise application design since the ascendancy of relational databases in the ‘80s will emerge as the result.
This article is published as part of the IDG Contributor Network. Want to Join?