Them bots, they are here. Are you ready?

RPA is one of the many “next big things” that the digital (erstwhile IT) world has come up with. The good thing is it adds real value to business, reducing cost, improving efficiency and catalysing process improvement. Most importantly, it does the “hard work”, letting human beings focus on the “smart work”.

Here are a few things you should keep in mind when you go in for an RPA implementation:

1. While the Big 3 tools are great options (UIP, AA, BP), there are others who can do the job at lower cost and are easier to deal with. A key issue with the Big 3 is that you may struggle with their implementation partners, many of who are boutiques with oodles of attitude grown at the altar of easy dollars from American and European clients. Indian market is not a priority for them.

2. RPA sales & pre-sales people have a good view of what the bot can do but are not necessarily the best persons to decide what are the all the things you can do with the bot. Remember this, the Bot can do anything that a person sitting at a computer and executing manually is doing without involving discretion. So, feel free to let your imagination run wild and figure out what are the manual process steps that a bot can do instead of a human being. You can always try out your ideas in a pilot / POC and figure out how to make it work before investing big money, time and effort into it.

3. Review the manual processes that are to be migrated to the Bot, specifically looking for areas where human judgement is used. The process may at the face of it seem like it is structured and easily automated, there are times when the person executing the step is exercising some judgment. This typically stems from either legacy process issues and / or poor data models. There are 2 ways of dealing with this. Either you build in some intelligence into the bot which can become a complicated, time consuming and expensive affair, or, you revisit the process and ensure that all implicit, unstated steps are codified or distilled into specific manual steps. This may involve tweaks at the process and data model level. Start with simple processes and bots and add complexity as you make progress.

4. This is also a good time to fix the process and data, getting rid of legacy inefficiencies and workarounds, and set up a streamlined process. Ensure that the inputs that are sent to the Bot are exactly the same every time, or, ensure that the Bot has checks built in to detect anomalies and either make the correction or send out an alert to the process owner.

5. There will be scenarios where the Bot stops working during execution. This could be due a problem with the systems that it is interacting with or because it encounters inputs inconsistent with what it has been programmed to do. In either scenario, you need to have a well-defined incident / exception management process.

6. Lastly ensure that the right security protocols are in place. This becomes more important if you are working with an external service provider and your systems are being accessed from outside your network. Also focus on minimising the possibility of the bot getting infected by malware or getting corrupted. This could be native or through its interactions with input and output systems. Either scenario can have significant repercussions for your business and must be protected against.

Questions? Write to me at arun@arcainnovation.com