OneReach.ai founder Robb Wilson is a UX pioneer with more than 20 years of experience working in design and technology. He has many useful thoughts about many things—conversational AI in particular. Some of those thoughts have been collected in his book from Wiley, Age of Invisible Machines: A Practical Guide to Growing a Hyperautomated Ecosystem of Intelligent Digital Workers. Others are shared here:

Few would argue that conversational AI is impacting businesses of every type. Unfortunately, most business leaders and organizations that have jumped into conversational AI have smacked their heads on a low ceiling dropped by the various limitations in the marketplace. Too many solutions focus on narrow elements of leveraging conversational AI—like NLU/NLP, text-to-speech (TTS or “read aloud” technology), automated speech recognition (ASR), and pre-built use cases. This leads to collections of disparate bots that can’t communicate with one another and are continually bumping into silo walls—unintelligent AI, if you will.

Using conversational AI intelligently requires a shared communication layer connecting people, systems, and things. This creates a fertile environment for sequencing powerful technologies in new ways, with people across an entire organization using code-free creation tools to continually build and refine automations. Gartner calls this “hyperautomation” and views it as an inevitable market state. Ready or not, hyperautomation is something every organization will have to reckon with, and they will need to focus on these requirements to get it right.

As Oliver Pickup wrote recently for Raconteur:

As an example of what’s possible with hyperautomation, take credit broker Loan.co.uk. The business, which has been building intelligent systems since 2014, has transformed mortgage lending from a process that’s traditionally been opaque, complex and painfully slow. The total automation improvements to date have “saved our 40 advisers and processors on average three hours and 45 minutes a day”, reports CEO Paul McGerrigan. The company’s AI helper, Albot, can search thousands of lenders’ offers in less than a second while matching more than 10,000 criteria, delivering the lowest rate appropriate for the applicant’s circumstances.

This kind of orchestration can provide better-than-human experiences, with technology running multiple concurrent processes that can allow people to make better use of their time. For organizations that aren’t already tech-first, the road to getting there likely won’t be pretty. It will require implementing a vast and exhaustive strategy for reorganizing internal technology and data, as well as a commitment to building automations right away. These early automations will fail, as will many of their lackluster successors, but companies that get into the rhythm of rapidly iterating and building successes off of the backs of these failures can put themselves in a position to succeed.