Enterprises continue to invest in core knowledge systems for functions across the business. Often, this takes a lot of preparation by a big team of data analysts, engineers, and scientists. The end result is often a fragmented technology experience and multiple sources of truth.
In this model, employees ask questions of their knowledge systems, and often receive outdated answers. Or, they scour through thousands of documents to no avail. The experience is painful. Despite all of the technology purchased in the last decade, this is the state of affairs for organizations: knowledge systems are ineffective, and humans are left without answers.
The future is not waiting for companies to get their act together. With the rise of Generative AI, pressure is mounting to connect LLMs into knowledge workflows to unleash new levels of efficiency and productivity – all with a simple natural language prompt.
The future is here with digital brains
Soon, companies will be able to run entire knowledge processes by typing a natural language question into an AI chat interface - just like they were searching for something on Google.
It won't be one monolithic model driving these outcomes, nor will they be completely autonomous. Instead, these processes will be dynamically created by “AI brains,” keep humans in the loop, and operate across a dynamic collection of applications, data sources and human experts. Each will launch their own integrated workflows or generate insights based on the very latest knowledge.
This is not a prediction - it’s a current reality. A team of Workato employees built it using Workato and OpenAI Functions. I watched a demo of this in a security context last week. The person typed a query “Bill said his laptop is acting funny after plugging in a USB drive,” and AI decided on the best possible series of steps across 6 different SaaS products, each time coming back to the person and confirming whether it should proceed to the next step with a rich explanation for why, and a confidence score. The original query was simple, but the end result was a complex, automated process that drew on company knowledge and mitigated a security threat: malware was identified, a laptop was quarantined, and a new laptop was provisioned and shipped to the employee.
These types of Generative AI use cases are on the horizon. We can envision them making the employee experience with technology more effective.
The importance of human-in-the-loop
In our CEO’s bestselling book The New Automation Mindset, three types of automation are described: task automation, straight-through processing, and orchestration. Many people stereotype automation as synonyms for task automation or straight-through processing, where everything happens automatically. However, orchestration is the ultimate form of automation, because it includes every element - people, processes, data, technology - operating in harmony.
Human-in-the-loop is a key part of this paradigm. And it has never been more important than now, as GenAI is getting infused into process flows. Our customers automate many processes including AI - and every one of these use cases features a human in the loop step. And rightly so. LLMs have not reached the place where they have earned enough trust to be set loose without human supervision.
Just like LLMs should get infused into orchestrations across applications and data, equally so do human validation steps need to be incorporated at the right moment.
Evolving from ‘good enough’
Businesses are only running more software. And each new app can add additional complexity. If they are not connected, they will create new challenges. To address these challenges, companies may build ad hoc app to app integrations. Despite the operational risk of such “hairball” integrations, for many companies, that’s good enough.
In the future with digital brains, “good enough” won’t cut it. In the new era of AI transformation, end-to-end process orchestrations are a must. Well architected data flows will be the lifeblood of these generative AI processes. Human-in-the-loop will supervise GenAI decisions. That’s not possible with home grown integration.
Companies have long-wanted to operate as one connected enterprise. Just adding new applications may have taken them farther from that goal. But now, with the combination of lowcode process orchestration including AI, companies have new motivation to break down the IT barriers that are hindering productivity and efficiency.
Combining automation, integration and AI into “on-the-fly” business processes holds immense promise. Businesses just have to make sure they are ready.
