Robots don't fail gracefully... by themselves.

October 19, 2020
Image of a team of people working in the office

The human experience is so critical to RPA success.

Technology professionals will be familiar with the concept of a ‘happy path’. If this is not a familiar term, in the context of software testing, a ‘happy path’ refers to the testing of a process reaching the likely outcome with few or no unexpected steps or errors.

It’s also best practice to prepare for the unexpected, or a ‘bad path, ‘sad path’ or ‘unhappy path’. At the end of the day, even the most experienced developer cannot possibly think of all potential outcomes for every automation. The idea of humans and robots working together, until now, has passed simple automation software tackling repetitive tasks and entered science fiction.

In this article, we will look at a few reasons why automation and bots may fail and how we, at Roots Automation, constantly work to overcome these obstacles so our bots never fail.

See also: What Did We Do Before Automation Technologies Like RPA?

Failure #1: System failures

Even after the due diligence of establishing the expected outcome and preparing for the worst-case scenario, the user can still find themselves experiencing underlying system failures or latency. It's only a matter of time before rules-based automation technology runs into an edge case and breaks.

Overcoming system failures

At Roots Automation, we train our bots to expect systems that will be slow or suddenly unresponsive. We teach the machine learning they run on how to quickly identify these issues, potentially work around them (where possible), and then let the appropriate parties know what is going on. Our team is also on hand to support, onboard your new bot (or ‘Digital Coworker’), and settle them into your team to ensure we can help provide the solution if challenges do arise.

We understand introducing new technology can be daunting and we are here to make this transition as smooth as possible.

Failure #2: Incorrect or missing data

Picture this, your new bot is paying a vendor’s invoice and it comes across a missing address. Without artificial intelligence, this could result in a pause in the overall process and a backlog of forms that did not get reviewed yet. This risk is an even larger issue in a financial services or insurance environment when the focus is on handling time-sensitive data.

Overcoming incorrect or missing data

We provide a Cockpit for our customers. The Cockpit is a dashboard whereby Digital Coworkers and humans can communicate. Through this platform, if a bot encounters an issue like this it will notify its human counterpart letting them know specifically what is missing. It will also anticipate the expected answer, e.g. if it is an address, the Digital Coworker will suggest the address by looking at past invoices from that vendor or going a google search.

Failure #3: Bots that don’t learn

To continue with the example above of the address, if a bot encounters the same empty address box error time and time again with no capacity to learn, the human counterpart may also find themselves doing repetitive work. This defeats the purpose of RPA software in the first place; to perform tasks at high-volumes.

Creating smart bots

We provide Digital Coworkers that self-learn. The ability to learn and improve over time is exactly why we call our bots ‘Coworkers’. If priorities change, our bots adapt and take direction from the humans, they can adjust to volumes and system performances, scaling automatically if needed to ensure the work always meets the SLA.

Our mission is to empower companies to free their ‘roots’ - their people - to focus on meaningful, impactful work like customer service and retention. We have created our unique Cockpit with the human user experience in mind. We strive to create the ideal coworker, with the capacity to learn and develop as a human would and to complete work with the extreme efficiency of a bot.

Reach out to us today to find the perfect Digital Coworker to join your team:

See also: Can You Rely on Automated Underwriting Systems to Approve Loans?

Demo a Digital Coworker On-Demand

Artificial Intelligence and automation can be difficult to conceptualize.

Access our on-demand demo and watch as a Digital Coworker learns to underwrite an insurance policy, eventually reaching a 95% straight-through processing rate.
Take me to the on-demand demo
Where is the value in process mining? A conversation between Roots Automation co-founders John Cottongim and Chaz Perera
Read More >
How to Overcome Five Challenges for Claims Processing in 2022
Read More >
Roots Automation is Excited to Announce a Customer Win
Read More >

Subscribe to our blog