Generative AI led by ChatGPT stirred the general public’s interest in the technologies surrounding hyperautomation, sending the race toward adoption into overdrive. This has added to a glut of point solutions to the marketplace that promise to deliver on the power of these technologies. There are the GPT-enabled tools popping up like crabgrass. There are also more sophisticated toolsets. Google Workspace is loaded with generative AI to help users write emails, summarize content, and create presentations (which it can also generate imagery and video for). These capabilities edge closer to using AI as more of a full-fledged solution, but there’s still a ton of ground to cover.

The main deficiency is in the ability to complete real tasks. Google Workspace seems poised to cut a lot of tedium from typical productivity models, but how much work can it do outside of its own box? Next-level applications of AI employ code-free creation tools to bring outside technologies and data sources into automated workflows that use GPT (or other NLP/NLU) as a conversational interface.

Large language models like GPT can explain complex things (like what snippets of code are meant to do) but they have a very limited sense of time and place and lack the agency to do stuff. I can ask ChatGPT to draft an article for me, but I can’t ask it to publish the article on my blog at 6am PST and send a summary of the article in an email to my subscribers an hour later.

If you want to use generative AI to hyperautomate you need to be able to complete real tasks. You need to find ways to elevate individuals and teams across your organization. No matter how stunning generative AI might seem, your current technology environment isn’t suited to maximize its true potential. You’ll need to find or build these critical components.

Critical Components of Platforms for Hyperautomation

A Contextual Memory System

The next big leap with generative AI won’t be improvements in predictive power, it will come with context. This context can’t come from NLU tools alone. A contextual memory system collects data from every conversation within an organization, across all channels, leveraging structured and unstructured data. This makes it possible to create channel- and user-specific experiences. Within this system, LLMs allow for rapid analysis of unstructured data, like emails, text messages, and recorded conversations. GraphDB or relational databases establish the relationships between data points, utilizing context that may not be assigned to the specific user, but is found in related datasets.

A Cognitive Orchestration Engine

Not all cognitive services are equal, and the pace of change is so fast that placing a bet on a single vendor guarantees suboptimal performance. I’ve also heard this referred to as cognitive architecture, but a cognitive orchestration engine can design experiences using both legacy systems and new market-best solutions. To create real, high-functioning automations, it’s critical that you can amalgamate language services (e.g. NLU, TTS, ASR, and localization) with other cognitive services, like computer vision and generative AI. This allows organizations to add vendors, manage cognitive services, and use them in different combinations, all in one place.

Intelligent Communication Fabric

It’s also crucial to enable the sharing of context and session information across channels and time. Gartner’s CX CORE report states, “Intelligent coordination is a form of human and technology orchestration, where customer relationship understanding and empathy principles prescribe a unique set of coordinated actions to be executed across an organization, resulting in a frictionless and relevant CX.”

Hyperglue

More things are broken rather than not inside most organizations. Technology isn’t working in ways that are even remotely close to what hyperautomation requires. If organizations can use ICF to standardize communications, a cognitive orchestration engine to enable legacy systems to collaborate with new technologies, and a contextual memory system to contextualize automated experiences, they can begin to repair.

Want to know more? These ideas have been expanded on in this informative booklet by our co-founder and CEO, Robb Wilson. Download here.