Historically, automation hasn't lived up to its supposed potential. Why?
- Too difficult to implement across disparate systems.
- Unfriendly for business users.
- We've only automated the obvious, and even those have taken a looong time to sort (e.g. accurate/automated expense processing).
- We seem to start from scratch each time. Why aren't we adapting automations quickly - from an apps store type of environment?
- Business users/domain experts often have excellent automation ideas, but lacked the tools to implement them (automation bottleneck).
Why is automation changing? My criteria
However, I believe this is changing. Why? No, not AI - though AI can definitely boost these efforts. Instead, I point to the convergence of automation and process mining - to surface workflow possibilities. And: automation vendors are getting better at putting business users in the driver's seat, without hanging out at the IT help desk, or submitting automation requests to ERP vendors.
As automation becomes more advanced, we need to ask harder questions. Here's the top items I am pressing with automation vendors:
- The "automation paradox" - where do we find the talent to create and manage automation at scale?
- The "junior associate" test - can a junior-level employee identify (or create) an automation prototype?
- The AI impact - can we provide smart, role-based automation recommendations to business users? Can we keep humans in the loop where needed, to validate process flows?
- Can we automate at scale, and manage those automations via an apps store type of environment?
- Can we automate processes that we never could before - beyond the basics?
- Where are the automation maturity models, to help customers learn from peers - and plan their roadmaps?
- Can we integrate automation capabilities into the users' preferred software environments, rather than forcing them into a new interface?
The automation skills paradox - can we democratize automation?
I don't believe any automation vendor has answers to all of these questions, but we might as well find out. Next up? Automation Anywhere. Here's part one: Have we entered the Automation Economy, or is intelligent automation just buzzword bingo? Automation Anywhere's CEO on what's next.
I've been thinking about the "automation paradox" for a couple of years now. But it was Avantra's John Appleby that crystallized this concept for me, during a fall 2022 Avantra Summit keynote:
There is so much work to do, and not enough people to do it. So you need to be thinking about, instead of being operators, they're automators. That is the talent transformation that has to happen in our industry.
And this is what I call the automation paradox. This is the biggest challenge of automation, that your key resources that have all the knowledge of all the stuff that you want to automate, are the busiest people in the business. And the simple fact is you have to get over that. You have to get over the hump, because if you don't, those people will never be freed up. They will be busy for the rest of their working lives. (Avantra CEO on why companies need to find a way through the 'automation paradox' ).
So I asked Automation Anywhere CEO Mihir Shukla: how do they handle this? Do they identify point people within customer organizations that become, essentially, an automation orchestration specialist? What does he think about the automation skills paradox? Shukla responded:
One thing that's important for people to understand is that in the old automation paradigm, this was always true, because automation was hard. It was designed for very sophisticated people with very high degrees of technical proficiency. That paradigm no longer holds true.
Yes, Shukla acknowledges, it's good to have a few automation specialists (In classic software lingo, I call these super-users). But Shukla says the "power of democratizing automation" changes that:
Let's say you are a bank, and you need to create an automated process to open an account... We have an idea called citizen developers that is well-governed. It empowers people to create a process. So somebody who is very tuned to opening a bank account process can easily create a process, and through various compliance governance checks - which we have also automated to the extent possible - you can then publish that, and make it available to 60,000 employees in the push of a button.
The fact that all of that can happen in 48 hours or 96 hours - that's not how people traditionally thought of automation. They thought of it as three/six months to a year projects, or even longer.
Automation in action - the Automation Anywhere example
But what does this look like in action? After our demo, Automation Anywhere sent me a collection of screen shots. I asked them to focus on the heterogenous enterprise. I wanted to see how people could use Automation Anywhere's tools - including their " automation co-pilot" AARI, while working out of their enterprise applications of choice.
We started with a scenario that showed how two employees could work together on a process workflow - with one running Genesys Contact Center, and the other in SAP. Via AARI, each worker could see the Automation Anywhere workflows embedded in their SAP or Genesys screens. In the hypothetical case I was shown, a customer is calling in about a claims issue - but the claims details are in another software system. AARI pulls those details onto the screen, without leaving Genesys:
The claim details are brought directly on the screen, so customers don't have to wait. Now we can focus on delivering a better experience. Immediately.
If the claims issue is due to a faulty error code, this can be escalated from the same screen (navigating through multiple screens and applications while the customer is on hold is a guaranteed loser experience). Instead, the call center employee connects to the claims specialist, working inside SAP. The claims specialist receives an urgent claims notification:
Naturally, she needs more information to make an informed claims decision. In the past, she would have had to search across three different applications. It would have taken her several hours to get that information collated.
For another embedded example, we switched to a Salesforce-based scenario:
The above screen is from a loan pre-approval scenario, via a hypothetical user working in the Salesforce Financial Services Cloud. In this case, the goal is to process loan applications more efficiently - a process that remains undeniably tedious for many applicants. Here is what is going on:
"Bria" sees a new loan application. This is where she starts to work directly with AARI directly in Salesforce. She sees a new loan application, and she'll run the document verification process behind the scenes. Document automation extracts and verifies the document, saving Bria enormous amounts of time.
Ahh, but here is where the AI aspects need to be dealt with. Today's AI systems, however useful they might be, are not 100% infallible. A well-designed AI workflow retains human oversight where needed - and learns from its missteps:
Now she has a chance to actually fix and perfect some of the results. Not everything was captured; the AI engine basically wasn't able to process a field. So Bria has to manually enter the information directly within Automation Anywhere. Document automation actually gets smarter every time she does this, because she's actually teaching and improving the AI model. So moving forward, tasks like this can also be automated.
Effective process automation should have role-based screens for automation specialists and managers. Here, we have a CoE Manager screen from Automation Anywhere:
This CoE screen provides access to a range of data, from idea submissions to ROI measurements, to governance and access control. One of the submitted ideas pertains to predictive supply chain planning:
Here we see an Automation Anywhere screen focusing on supply chain cost reduction scenarios, with recommended processes on the right hand side, weighted according to projected ROI.
Where does automation go from here?
But if companies want to advance into automation maturity, they need to figure out how to scale their automations - and make the tools more accessible to more users. So I asked Shukla about the customers on the advanced side of this - what are their characteristics? Obviously scale is clearly one of them. Is there anything else?
Some of them have set a goal: 'I want to automate 10% or 20% of all my work in the next three to five years.' Especially for a large organization of 60,000 people or more, that's a lot of work. But they're strategically approaching it and saying, 'Look, I'm going to fundamentally change how I operate.'
But how do you measure the progress?
The way they're measuring it is: let's say I have 100,000 people, and they work x hours a year. I know exactly how many hours of work people are doing. 20% of it has to get automated or 10%, depending on where you set your goal. The second goal they are setting is: '60% of my knowledge workers must work with a bot in the next few years.'
But for Shukla, you can't get there on ROI alone - the change goes deeper.
It's a culture change. I might save $200 million, but if it's limited to 100 people, I'm not changing how I think and how I work in the future. By allowing this to go across the organization, it becomes a way of thinking and working for everybody.
There's one more criteria of advanced automation: automating things you've never automated before. Shukla:
In the last COVID years, we all suffered through supply chain challenges. There was a customer who had 22 warehouses, and 365,000 items. Previously, if they ran out of an item, the warehouse would just order more. But there weren't any more to come during COVID.
So they wrote a bot that runs four times a day. It compares inventory levels at all warehouses, across all items. If one has ten and the other has zero, and zero has a demand, it will move the item between warehouses, to save hundreds of millions of dollars in inventory cost. Now, this is an example of a bot. This is not something you can do manually.
To address all the advanced automation criteria I listed, there is plenty of work ahead. I give Automation Anywhere credit for not presenting me with totally seamless demos, or sugarcoating the change that lies ahead:
What we're seeing is the ability to scale automations for a lot of organizations is still tough, whether it's through the complex, existing technology ecosystem they have, or how to arm the rest of their organization to have the ability to deliver automations to meet the needs of the business.
Yes, that is the straight talk we need. So how do you drive this - and achieve better adoption? Customer success programs - along with published maturity models - are both necessary. I didn't have the chance to get further into those with Automation Anywhere this time around, but they have programs in play. Their Automation Success Program, launched last fall, makes automations available across an organization. The Automation Pathfinder program is their answer to fleshing out those maturity models - and building a community around so-called "automation journeys."
Though I wasn't able to easily locate a public automation maturity model on the Automation Anywhere site, the Automation Pathfinder program presents useful role-based content "journeys," depending on where you land in your organization.
I appreciated that Automation Anywhere didn't overhype the power of AI beyond its constraints, acknowledging the possibility for AI errors, and presenting human-in-the-loop workflows. Those AI workflows still qualify as significantly improved, so why do so many vendors overstate what AI is capable of? That helps no one. I don't believe that any of this is easy. Silos of data - not to mention departmental politics - don't just magically go away because our automation tooling has improved.
To find the organizational will to change, you need inspiration to get you through the slog days - and there will always be slog days. Automation the mundane is non-negotiable - but it's not inspiring either. Shukla has a different rallying cry:
Initially, we are automating things that we were doing before. But if you look at our usage of the mobile device, we started doing things that we never did before. That is the true purpose of technology. You are going to begin to re-imagine work - and start doing things that you never did before. And that is the true transformer.
Now that's a worthy goal.