Artificial intelligencce (AI) alone is not enough to drive true business value from autonomous operations
Since the mid-1990s, business processes have made a series of rapid evolutionary leaps towards greater automation and efficiency. Another evolutionary leap is happening right now. One that will, in time, usher in the age of autonomous operations, eventually leading to (semi) autonomous enterprises.
If you think this leap is all about AI, well, your business could be heading towards an evolutionary dead end. I discussed why that’s the case in a recent webinar with my old friend and the godfather of process mining, Professor Wil van der Aalst. I’ll share some of his wisdom here, but first, it’s important to put this evolutionary leap in context.
The evolution of business processes (a quick history)
Leap one
Businesses computerize their paper-based processes. As most of you will remember, this happens circa the tail end of the last century. The same people manage the processes, and make the same mistakes, but they use enterprise resource planning (ERP) and customer relationship management (CRM) instead of sticky notes and frantic shouts across the office. Many businesses also embrace workflow management systems in an attempt to model and automate their processes. These attempts tend to fail, because their models don’t reflect their processes’ messy reality.
Leap two
Process mining arrives (thanks, Wil) and lets companies model their processes based on event data from those computerized systems. Suddenly, it’s easier to understand how your processes actually work and businesses are able to get back on the automation train.
Leap three
Robotic process automation (RPA) allows discrete, mundane, repetitive tasks to be handed to tireless bots.
Leap four
Someone, somewhere, takes an important step back. They say, “Sure, we can automate this specific task, But what’s the desired outcome of this whole process? How do we optimize that?”
Thus begins the leap from automated processes to orchestrated processes. It’s made possible by the development of object-centric process mining, which, when combined with the right algorithms and process knowledge, gives rise to process intelligence and the ability to create process digital twins.
However, we’ve still not reached process perfection
Orchestration, as the name suggests, finally lets businesses direct their processes like different players in an orchestra and create one, grand symphony of efficiency, but it’s not the endgame for process optimization.
Even beautifully orchestrated processes can harbor inefficiency. In fact, when a business evolves to process orchestration, it sees just how many decisions are still being made by humans that could be made, more efficiently and accurately, by machines: decisions based on patterns and rules, like whether to remove a credit block on a customer’s account.
In some businesses, AI is already making decisions like these, while humans focus on handling the exceptional cases. This represents the next big evolutionary leap, the jump from orchestrated processes to autonomous processes. It’s a leap that AI, on its own, simply can’t deliver.
Why AI isn’t enough
Remember how, when ChatGPT first launched, it was useless at mathematics? It’s only better now because, instead of guessing the answer, it puts the question to a calculator.
Similarly, if you ask ChatGPT how to overcome a production bottleneck in your business, it’ll guess, offering generic recommendations that may, or may not, be appropriate.
What if you give a generative AI tool access to the engine your business is using to mine and orchestrate its processes? If it teams AI up with process intelligence? Then, like ChatGPT wielding a calculator, it can provide answers that are informed, meaningful and useful.
Talking to your processes
With the right AI copilot you can now ask questions like, “what’s my on-time delivery rate?” and receive a perfect answer, calculated based on your business’ object-centric event data. It’s like having a fully informed consultant, sitting in front of you, at all times.
When a business develops these capabilities, it’s also unlocking that next evolutionary leap: from orchestrated to autonomous. After all, when your AI tools have the context and data to make good decisions, you can start trusting them to take the reins.
As Wil put it during our conversation, “AI is a wonderful technology, but it needs context, and if you very naively think that you can use AI without having your data management – let’s say, organized properly – that’s going to fail.”
Towards autonomous operations and enterprises
You can watch the webinar on demand and there’s much more insight from Wil to discover. He dives into everything from the types of AI businesses are sleeping on, to the way work will be redistributed as we progress towards fully autonomous enterprises.
Wil and I agree that this will be a gradual journey, just like the one towards self-driving cars, but we agree on something else, too. It’s a journey that smart businesses are starting today.