5 intelligent automation lessons learned: Microsoft’s Rohit Amberker



Seth Adler
06/18/2018

The AI & Intelligent Automation Network asks key practitioners for five lessons learned from their intelligent automation journey

Next up is Rohit Amberker, director royalties and content operations at Microsoft.

I’ve taken the opportunity to provide a bit of additional content to the list. Amberker shares top 5 lessons learned from his intelligent automation journey:

1. RA: Stabilizing processes is critical before thinking about automation. As Bill Gates said, "Automation applied to an efficient operation will magnify the efficiency. Automation applied to an inefficient operation will magnify the inefficiency."

SA: You may have also heard s#!% in, s#!% out. When AIIA moderate roundtables at events it surprises me that executives are still keen to discuss automation prior to ensuring that targeted processes are efficient. Some of the RPA failures are of course do to the fact that you’ve got to clean up the garage before you start practicing with the band.

Read more: Top 5 lessons learned from intelligent automation: Cindy Gallagher

2. RA: For early value realization, it's best to apply automation on the "happy path" i.e. predictable, stable and exception free processes. In any spread of events / transactions normally distributed, large population will fall within one standard deviation or the "happy path."

SA: There’s a bit of a debate on this one. One philosophy is making sure you get a win, then adjusting and then scaling-up. Another philosophy is if you don’t start with an end-to-end process, you’re not going to know what “efficient” means. You’re also not going to truly know if you can actually scale-up.

3. RA: Things break all the time. Absolute robustness is an illusion. Building and periodically testing 'resiliency' in processes / solutions should be a core organizational priority.

SA: Unpacking this leads to a few thoughts… The first is yes, robots do ‘get sick,’ so take that out of the arsenal of points of discussion when talking about human vs. intelligent workers. The second is, how do they ‘get sick’? If you’re attempting to put your bot on an island so it succeeds, you’ll find that’s nearly impossible when someone changes something somewhere else which affects your bot island.

Read more: 5 lessons learned from intelligent automation: Jon Theurekauf

4. RA: Deploy ML/AI to automate controls and identify process exceptions real-time and before the event. A post facto sample based audit is not helpful anymore.

SA: Start with the end in mind. More on this as the industry catches-up on implementing ML and AI solutions…

5. RA: Reskill people to focus on customer empathy and nurturing relationships. This is where carbon intelligence is still superior to silicon intelligence.

SA: Right yes, those saved FTE’s… don’t just redeploy willy-nilly, thoughtfully place your carbon-based workforce—formerly known as people—into roles that involve the customer. Humanize your enterprise though intelligent automation.

OUR BENEFACTORS