3 Ways Computer Vision and Granularity Beat Traditional Process Discovery
This article was originally printed on the FortressIQ website and includes updates to reflect the company’s change in status. FortressIQ is now part of Automation Anywhere, and its offering is now our process intelligence product FortressIQ.
Processes are the lifeblood of your business. They move people, goods, knowledge, and money around your business, keeping it afloat and moving it forward. But few organizations have a deep understanding of how those processes, and their business, work. Not only is it difficult to improve a process you don’t understand, it’s also difficult to determine where and what to improve to make a material and sustainable impact on your business.
Many organizations attempt to understand their processes using traditional process mapping, process mining, and/or process discovery. Each of these methods, however, fails to provide a complete view of how a business operates. Process mapping uses slow, expensive manual tracking and is limited to small sample sizes. Process mining uses software application logs, but misses tasks performed in common non-logged apps, like email and spreadsheets. Process discovery overcomes the log issue with technology but requires slow, expensive manual interpretation of the results.
Unfortunately, not one of these solutions provides the granularity you really need to transform business processes. Relying on these incomplete sources leads to missed nuance, subtasks, or variations, which then results in process optimization and automation failure or expensive rework. And the lack of granularity leads to missed opportunities or even more process improvement disappointment. The result, according to McKinsey & Company, is that 86% of organizations fail to deploy sustainable business transformations.
Why granularity matters
Today’s business tasks and processes flow across legacy software systems, modern cloud-based applications, and common email, spreadsheet, and chat productivity tools. A simple purchase order, for example, might be initiated with an email, require inventory data from a legacy system, utilize part numbers copied from a spreadsheet, and be eventually generated on a cloud-based app through a web browser with data pasted from the spreadsheet.
If you’re standing over someone’s shoulder, it’s easy to see how the process unfolds, which applications are used, and in what order. But now consider how different workers or teams or departments might have slight differences in their process. Maybe there are approvals required for any purchase over $500 in Legal, but Marketing’s approval threshold is $1,000. Sales might initiate purchase orders in Salesforce Chatter while Operations uses a shared Excel spreadsheet, but only Shipping requires workers to enter detailed information about the purchase. That’s a simple example, but if you made decisions on just a sample of manually captured process data, you would miss some incredibly important subtleties across the business.
Even seemingly minor differences can derail process optimization or automation initiatives. Knowing how many POs are created, the average process time, and even the likely tools involved is not nearly enough information to make cross-enterprise decisions on process improvements. And you’re not going to stand over the shoulder of every worker to note myriad subtleties across your business. But software can.
How computer vision captures process granularity
Computer vision is simply giving computers the power to see what you would see when looking over a worker’s shoulder. It’s a complex and advanced field of artificial intelligence (AI) that allows a computer to “see” important elements in an image such as people, numbers, words, or anything else. Think of how video chat uses effects to put digital sunglasses on your face or a funny hat on your head. The computer in your smartphone sees your face and its movements and matches the placement of the digital item.
The Automation Anywhere FortressIQ Process Intelligence solution uses software-based sensors that are deployed on a worker’s desktop or laptop to continuously capture images of all screen activity. Computer vision then converts those virtual over-the-shoulder snapshots into a detailed blueprint of every process across every application and department, with universal compatibility and without painful integrations. It doesn’t matter if a worker uses email, chat, a browser, a virtual machine, a legacy app, a native desktop app, or anything else; FortressIQ captures the necessary process steps with deep granularity.
With computer vision and machine learning, the FortressIQ solution sees the screen as a human would but doesn’t disrupt workers. And it does so securely and at scale, for every worker across your enterprise, in real-time. It has the speed and coverage to capture, record, and analyze granular steps in complex use cases, plus uses AI to quickly identify new opportunities.
Benefits of granular process intelligence
FortressIQ captures process data so comprehensive (broad) and granular (deep) that, after just a few weeks, you’ll discover the detailed process insights necessary to answer the most pressing questions facing your organization. Computer vision provides the granularity and scale to accomplish change, automations, and optimizations unattainable with traditional process discovery methods. Here are three ways it beats traditional process discovery.
#1 – Discover application usage opportunities
FortressIQ provides application usage details in both time duration and frequency, down to the individual user and application level. This can help you quickly understand worker focus, how to improve training, where process friction is present, and if applications or IT issues might be causing problems.
For example, FortressIQ can detail the time spent on each specific application over the course of a specific process. When building a process optimization queue, you can focus on the applications where workers spend the bulk of their time to maximize the impact of any changes.
#2 – Determine the impact of process optimizations
FortressIQ automatically provides granularity down to the specific screens used within each application, by which workers, for which tasks, and for how long. Then, when optimizing an end-to-end process, you’ll know where the most impact and largest improvements can be made. Without this granularity, you wouldn’t know where bottlenecks specifically reside and could end up focusing on the wrong or unimpactful portions of the process.
Since FortressIQ produces data at the screen level, you can pinpoint and prioritize which tasks and processes are the best candidates for automation and optimization.
#3 – Separate reality from assumptions
FortressIQ uses computer vision to automatically determine the screens and sequences used in a process and then uses machine learning to determine which controls are used. The solution can then automatically generate process definition documents (PDD) at the task and event-level with accompanying screenshots for each step. This level of granularity shows the exact text boxes, drop-down boxes, and application screens being used at every step.
Intelligently generated PDDs based on comprehensive quantitative data, not just samples or interviews, show you exactly what’s happening in your business. The PDDs can then be used to define automations, improve training, and guide optimization efforts since you’re working with what happens versus what you assume is happening. A common example is finding workers spend a lot of time in a text box on a particular screen when you can speed the process and drive standardization by providing a drop-down box with common entry values.
Computer vision and granularity for the win
Granularity eliminates guesswork to show you exactly what’s happening in your business. Computer vision captures that granularity but uses AI to do so incredibly fast and at scale. The resulting accuracy and comprehensiveness empower you to rapidly transform your business, knowing your process automations and optimizations will have a lasting impact.