Looking at the world today, we live in an endless series of processes. There’s a process to buy a car, a process to check out at the grocery store, and processes to make appointments or pay bills. Behind all of these are “business processes” that are oftentimes misunderstood or not understood at all.
Within Fortune 1000 companies, there’s an assumption that business processes are well understood. This turns out to often not be the case, as there is an entire nascent industry dedicated to process discovery and process intelligence. On the other hand, there are well-established workflow vendors that purport to have easy-to-use platforms for business users to construct a workflow or business process with no involvement from IT. But realistically, is it possible for a business user to design a process that may touch tens or hundreds of (at times, highly technical) backend systems? Further, would they have the information necessary to know what to create if given the opportunity?
The elusive “business process” is a bit of a misnomer. Business processes rely on humans, and humans are fallible regardless of how efficient and well-performing they can be. No matter how hard we work, we can never be “perfect”—and that statement grows as the volume of work increases. That has led top organizations starting to qualify automation to increase throughput and achieve cost savings, but how can you ensure you’re investing in the right solution?
Let’s start with process understanding and a generic process, like order to cash. Every business has this process regardless of whether it’s a car dealer, retail store, or financial services institution. It’s the lifeblood of the business (afterall, it’s essential to know how much you’re spending and receiving), but it’s often extremely complex, and in large organizations, it can touch hundreds of existing systems. So how do we dissect this process and use that knowledge to improve it?
Traditional process mapping techniques encourage organizations to start with an as-is process. Process discovery software is a new market and results may vary, but generally through some combination of software and interviews, you can get a rough idea of your process. That’s step 1.
Step 2, if you’re using a traditional workflow tool, is comparing vendors that are telling you that their tools are easy enough for a business user to design, build, and implement.
“Drag and drop!” “Connect with arrows!” “Business-user friendly!” In reality, however, some amount of code will have to be written, and developers have to get involved and write the necessary code to make this work.
In our order to cash (or OTC or O2C) example, API calls need to be made, documents need to get routed to different queues, units of work need to be routed to different people across different departments. Maybe a message or document needs to get sent to an external vendor, in which case we’re bringing IT & security teams to the table to ensure that connection is robust enough to meet information security guidelines. Some of this is easy enough for a business user to enter into a well-designed interface, but depending on the size and functionality of the organization, complexity and variability are ripe with risk.
Putting that aside, the process is now built and deployed to production. In our order to cash example, our customers are placing orders, and we are collecting payment. All is well. Until it’s not. Joe in Accounts Receivable figures out a way to speed up his part of the process because he finds a field that he feels is not necessary, skipping this data entry point. Sue in Accounts Payable finds it’s easier to send a Slack message to a colleague to get an account number rather than perform a lookup in a database herself. And so on and so forth. Pretty soon the idealized business process breaks down.
Now let’s play this forward. Leadership at this fictitious company sees something is off in the process and decides to implement Robotic Process Automation (RPA) software to speed up the process and minimize human work. Developers build bots to solve for specific spots in the process, and things speed up for a short time, but when a third-party software changes their UI, the bots break because they’re unable to adapt to change. Time to call the vendor and ask for an update (can you hear the hold music already?)
What if there was a better way to implement this order to cash process? Something flexible and fully-configurable that meets the needs of your current business process, while being adaptable for the future of your organization’s needs?
There’s a desired need to find a single solution that can be a panacea, and typically this is a workflow engine. Going back a few years, the concept of process orchestration, where a piece of software could orchestrate a process but not execute it, was popular, but this has fallen out of favor recently as these legacy systems and applications failed to keep up with the changing needs and requirements of enterprise organizations.
So what options are left to build an efficient, reliable, and future-flexible business process? An intelligent automation platform that utilizes Artificial Intelligence (AI) and Machine Learning (ML)?
AI and ML have solved for some of the more rigid, rule-based challenges that automation technologies like RPA have historically fallen victim to. With these technologies at the heart of the Hyperscience Platform, we’ve developed a solution that is able to process customer data with variability in mind.
Historically, our automation platform has consistently excelled with extracting, enriching, and validating handwritten text from structured documents, such as a W-2 or W-9 form. Our solution can also handle semi-structured documents, like an invoice or check, and it can even read my chicken scratch handwriting!
But with the latest release of the Hyperscience Platform, we’re moving beyond semi-structured classification and extraction into document-centric flows. This is not meant to be a replacement for a workflow or process orchestration engine, but rather a fit for purpose tool to ingest documents from all inputs (whether handwritten or electronic), to perform validation checks and other document input- and output-centric tasks, and then to send structured data downstream for faster, more reliable processing.
We haven’t ignored business users either! With the Hyperscience Platform, business users can tweak settings, create rules, or set various thresholds when it comes to extraction abilities. With this approach, not only do you get a fit for purpose platform, but it’s also designed to engage the correct people in your organization at the correct time and co-exists with other solutions that may be in place already.
One person can’t design a thoughtful, optimized, long-term business process using a single tool, nor should they.
While other solutions are designed with an “either or” mentality, such as business users or tech staff, fully automated or not automated at all, we’ve built an intelligent and flexible platform that enables machines to work hand-in-hand with their human counterparts. Rather than solving for singular issues or processes, we’ve designed a platform that easily adapts to the needs of each team and technology, both today and for tomorrow. After all, a “business process” should work for – and not in spite of – your organization.
Interested in seeing what’s possible with a solution that is built with this mentality? Request a demo today and learn a new way to envision a business-centric workflow.
Dan Rabinovitz is a Solutions Engineer at Hyperscience. Connect with him on LinkedIn.