Paving the Way to Success in AI and Intelligent Process Automation
While a lot of AI implementations are still exploratory these days, there are many examples of companies starting to gain real traction behind the scenes – applying AI to automate manual, back-office business processes.
Sometimes referred to as intelligent process automation (“IPA”), these back-office use cases focus on manual, document-based workflows such as contract analytics, audit planning and reporting, RFP analysis and composition, sales opportunity workflow automation, customer support analysis and automation, appraisal, and claims analysis.
These use cases are fundamentally different than RPA use cases in that they deal primarily with unstructured content, and rather than following simple task-based automation, they require some form of AI-based decisioning in order to augment or automate the workflow.
In our work with clients, typically large enterprises, we see a lot of different approaches to these projects. And there are some clear distinctions between those projects that are succeeding vs. those that tend to stall and fail. Here are some of the key lessons we’ve learned that you can take into account as you try to move your AI initiatives forward.
Start Small, with a Business Outcome In Mind
We see a lot of AI initiatives that begin as discovery projects without any real business outcome in mind. Almost all of them stall. While it’s absolutely worthwhile to experiment with AI to see what’s possible, it’s also very difficult to do that without a real use case to test it on.
Big organizations with very large data science teams might have this luxury, but the large majority of organizations don’t. Do you have a business problem you are trying to solve? Do you have a use case in mind? If not, it’s important to find one. It does not have to be big or strategic, but you need a defined problem to experiment on and learn from.
I suggest looking for low hanging fruit; for example, an opportunity to automate a manual business process that would make a difference for a defined group. Define some baseline metrics with the business at the start and track them along the way. These may include both hard costs and opportunity costs. Prove your project’s value within this defined scope, then expand on it from there.
Be Realistic About What AI Can and Can’t Do
When it comes to unstructured content, many organizations come into their project with the assumption that AI can tell them what the right answer is inside a large pool of data.
In reality, AI is great at discovering what maps to or matches an already defined desired state; e.g., if this is what compliant contract language looks like, AI can then automate the process of identifying which contracts are compliant. which are not, and then recommend next steps.
But, if you can’t define the desired outcome, don’t expect AI to do it for you.
Make Sure You Have Access to the Right Data
In my experience, you have to have data that can define your desired state. This may involve having your business SMEs label or annotate examples of what reflects the right answer vs. examples that represent the wrong answer. It does not necessarily have to be a lot of examples, but you do require very clear examples of right and wrong to put AI to work against a larger data set.
The other thing to keep in mind is that your internal enterprise data is much more digestible than trying to scrape data from across the Internet. In our experience, ‘boil the ocean’ type of AI projects are doomed to fail unless you are Google, Facebook, Amazon or Apple.
Match Up IT/Data Science Team Members with SMEs in the Business
If AI needs a business outcome, data science and IT teams can only go so far without people that have expertise about that business outcome or process. Too many AI projects get too far down the road before the business people get pulled in.
When this happens, it can be very difficult to get a project back on track. It’s important that both groups understand their essential roles in the process. We encourage companies to task their IT/data science staff to go out and work with the business to find use cases that might benefit from AI and intelligent process automation. As those efforts deliver value, The SMEs will then instinctively look for similar use cases where that value can be replicated. It’s also important to balance the demands on the SMEs’ time and the projected value to come of it. That should not be overlooked – as it could become another impediment to adoption.
Manage the Amount of Data Science You Expect from Your Business SMEs
This one is directly related to my last comment. There is a lot of talk about citizen data scientists and I believe that trend will become even more prevalent over time. But don’t expect those folks to understand algorithms and data models.
Instead, leverage them to help identify and label the right data inputs so that everyone can agree on what represents the right and wrong outcomes in the context of the business goal at hand.
Ensure a Common Understanding of the Business Process in Question
As I mentioned at the start of this article, we see AI being used a lot to automate existing, back-office business process; e.g., In most organizations, you’d expect these to be pretty well-defined processes. But we’re continually surprised at how different people in an organization view how a specific business process works.
It’s important to start with a common understanding of the different steps in a process before you can apply AI to automate it. Don’t make any assumptions. It may not be nearly as close as you think.
It’s also important to be patient. It’s still early days for the broad application of AI and intelligent process automation. Different organizations are finding traction with different business processes and use cases. Several iterations will likely be required to find the sweet spot in your organization. Use the scientific method, find out where you can get the most traction, and continue to build on the value you deliver along the way.
Starting Down the Path of Success with AI
To increase the likelihood of success for your AI and Intelligent Process Automation projects, you need three things at a minimum:
- An existing, defined process or workflow with a clearly defined desired outcome as a candidate for improvement, augmentation or automation.
- Representative data, content or documents for the given process.
- Input from the key Subject Matter Experts (SMEs) to define and evaluate success.
About the Author: Tom Wilde is the CEO of Indico Data Solutions, a provider of enterprise AI solutions for intelligent process automation. Before taking his current role, Tom was the Chief Product Officer at Cxense (see-sense). He also founded Ramp, an enterprise video content management company, and held senior roles at Fast Search, Miva Systems, and Lycos. He is a frequent industry contributor and earned his MBA in Entrepreneurial Management from Wharton.