Skip to main content

Are your bots working hard or hardly working?

Becky Blackwell, 6 minute read

I talk with a ton of RPA Centers of Excellence about how they manage their bot farms. We talk about their challenges – what they spend too much time and money on. We talk about their failures – the things they didn’t foresee and how they can prevent them in the future. And we talk about their successes – the ingenious things they do to make their teams more productive. A recurring theme has been a tendency to over-provision bot farms to be sure that there are always enough bots on hand to handle unexpected workload peaks.

Over-provisioning bots comes with a lot of overhead. You are paying for the VM. You are paying for all the software loaded on the VM. You are paying IT to configure, secure and update the VM. You are paying an admin to create schedules for the bot. All of this ensures that your people work hard while your bot farm hardly works because the majority of bots are only needed during peak times.

This is where Auto-balancing was born. Auto-balancing automatically reallocates bots based on SLAs and real-time workloads. It keeps them busy 100% of the time, working on tasks for which they are needed the most.

Some unattended robotic work has a high priority – it needs to get done quick. But other robotic work can slack a bit – it needs to get done by tomorrow.

With Auto-balancing, every work assignment has an SLA attached to it, and these SLAs can be different for each type of work. They can also be different based upon how the work came into the system. For example, the SLA would be long for work that can wait until tomorrow, like generating an end-of-month report. And it can be short for work that needs to be done quickly, like researching an account so that a ‘worker’ can respond to an email in a timely fashion. All robotic work assignments come into a robotic work group with an SLA attached. Pega Infinity has been handling SLA’s for years, so this part is easy!

The harder part is making sure the bots move around and get all this work done on time.

At Pega, our bots work on one work group at a time. Each work group contains one-to-many robotic work queues and the bots working in a work group are skilled to handle all work in those queues. This means that they have the required automations along with access to applications and the credentials needed to run those automations. As long as the bot has the required apps and creds, they can move between the work groups as needed (the automations are automatically downloaded when starting work on a new work group). So, one bot (which is usually one virtual machine depending on the loaded operating system), if properly skilled, can do any and every type of robotic work needed.

But how do you make those bots move around and get the work done on time?

The Auto-balancing engine analyzes the SLAs of all work in all queues. And it determines how much bot-force is required to get everything done on time. The engine stops bots that aren’t needed and moves / start bots where they are needed the most. This means that the most important work is handled during peak times and the less important work can be handled during quiet times. AUTOMATICALLY.

And now with Robot Manager 8.5.2, we have layered in an even better feature. This new feature enables our clients to prioritize work groups, giving the system the knowledge to know which SLAs are more important. So, if the system is saturated with too much robotic work and not enough robots, the Auto-balancing engine will know which work groups to prioritize and make sure that the most important SLAs are met.

Pega bots work hard!

To learn more about Auto-balancing, check out this Tech Talk. And click here to learn more about Pega RPA. Are you using Auto-balancing today or interested in learning more? Join the conversation on Collaboration Center!

 

Don't Forget

 

Did you find this content helpful?

We'd prefer it if you saw us at our best.

Pega Community has detected you are using a browser which may prevent you from experiencing the site as intended. To improve your experience, please update your browser.

Close Deprecation Notice