We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
This article was contributed by Sergio Suarez Jr., founder and CEO of TackleAI.
Workflow automation, also called data automation, is the new buzzword in the tech industry, but what is it, really? To put it simply, workflow automation means getting information into a system, launching tasks to process the data, and routing it to the correct people. This process is done using rule-based logic, and ideally with little to no human intervention necessary.
For some people, a world of automation and artificial intelligence can be overwhelming or even lead to job security concerns. In reality, automation and artificial intelligence can be complementary work companions to most jobs, helping humans do their jobs more efficiently and effectively.
One example of workflow automation is in the way most invoices are processed. The typical manual process is tedious, time-consuming and prone to errors. This process usually includes receiving an email, forwarding it to the billing department, and assigning it to someone who then types out the important information into a payable invoice. That information then gets sent to someone to take line items and do an inventory to confirm the information is correct. Each time this process involves another person, there is a greater chance of human error and definitely more time added to the process.
This is why automation, specifically related to documents like invoices, can be massively advantageous for organizations in terms of finance, efficiency and accuracy. Still, there are different approaches to workflow automation that pose different advantages to different organizations, and an informed decision between the two is critical.
There are two types of workflow automations – Intelligent and RPA (Robotic Process Automation). The data that these two automation types process may be structured or unstructured. RPA can only process structured data, meaning data that has labels or headers, and is staying in the same part of the page every time. Intelligent Workflow Automation, which is true artificial intelligence, can process, extract and classify unstructured data without ever having seen the document before.
In the example of the invoices: with RPA, a company will receive an email, scan it, and train the software for that particular invoice. This works well when the document contains structured data, with the same information in the same spot, and has neat labeling or headers, and the amount of space in line items does not change.
With Intelligent Workflow Automation (IWA), once it has learned one type of invoice, it now can read any type of invoice, and extract and classify information, such as the payer and payee’s account information, and know which one is which. With each new document, it not only processes the data and makes sense of it, but learns and trains itself to understand the next document faster and with more accuracy, each time. This is vital for an organization to save time and reduce error rates, as 85% of data is unstructured.
Both RPA and IWA offer advantages to organizations. In terms of capability, Intelligent Workflow Automation wins out, but in terms of startup cost and maintenance, RPA can provide a strong solution.
There are a few steps to implementing automation into your business infrastructure, and they differ slightly depending on individual objectives, industries and technological capabilities.
First, identify the task at hand: while a small sales firm might look to automate a standard, structured invoice, a hospital might need to process a variety of dynamic documents and EHR data. Depending on the complexity of the data, the sensitivity of the transferred information and the steps involved in each task, one can select either RPA or IWA as the preferred automation option. With complex tasks, IWA can go as far as completing redaction, or data extraction and validation, which are necessary in medicine, government and law.
Second, select your “provider”: Will you hire a full-time developer? Will you engage with a firm? Will you implement a pre-developed solution? These are all strong options that differ in value between budget, and again, the complexity of the task at hand.
Then, launch: getting automation right requires a lot of fine-tuning, so when the ball is rolling, make sure to pay attention. The beauty of the technology is its ability to continually learn and optimize itself, but much of this learning ability depends on the human team’s ability to feed it intelligent information.
A specific use case of implementing workflow automation occurred with a large healthcare data aggregator. Hospital systems sent the data aggregator hundreds of thousands of dynamically unstructured healthcare documents each day, which then needed to be classified, and have the data extracted and populated into each client’s electronic health record system. The workflow involved sending the documents overseas, to be manually read, classified and typed out into an Excel spreadsheet. This process took up to 48 hours to complete each time, and had an average error rate of 10-30%, along with potential security risks and HIPAA violations.
The hundreds of thousands of documents sent were all dynamically unstructured, so the healthcare data aggregator went with IWA instead of RPA for its automated workflow. The IWA was now completing the same workflow in less than a second. Data points were now extracted from all types of unstructured documents, such as prescriptions, MRIs, and oncology reports. The IWA is 70% less expensive than sending the same documents overseas for manual classification, and saved 50% of the operational budget. Now 73% of the documents are processed with IWA and need no human intervention at all. The error rate was reduced to only 0.03%, and the IWA can process documents it has never seen before.
In short, automation is more than a buzzword. It’s an opportunity to boost revenue, optimize performance and accuracy, and use immense technological power to allow human workers to maximize their potential outside repetitive and time-consuming data tasks. In 2022, automation is within reach for businesses and organizations of all sizes, unlocking opportunities for efficiency and revenue that weren’t previously available. A business with automation is a business with an advantage – and it’s never too late to get started.
Sergio Suarez Jr. is the founder and CEO of TackleAI.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!