Automation impact

End-to-End RPA explained
in simple words

Automation impact

End-to-End RPA explained
in simple words

AI3 – Operating model – Part 2

AI3 - Operating model - Part 2

Topic: Operating model
GuestFrank Schikora
Recommended for: CoE management, CFO, Team managers, RPA Lead

Operating model - Part 2

About guest: Frank Schikora is a Head of Delivery @ Roboyo and UiPath MVP

Episode content:
– 1:10        Importance of Project Champion
– 4:21        Where does the CoE start?
– 6:30        Skepticism about RPA
– 11:25      Implementation of RPA in Business vs IT
– 13:50      The Mixed Model (Hybrid Model)
– 16:38      Does the Operating Model deviate from business to IT?
– 19:10      Centralized, Federated and Decentralized CoE models
– 56:56      Mistakes and Myths when setting up Operating Model
– 58:43      Does Covid-19 affect the Operating Model?
– 1:02:20  The Operating Model 3.0
– 1:06:05  DOs and DON’Ts for successful Operating Model


Center of Excellence has shifted throughout the years. While in 2017 it was almost exclusively in business, now it is more often driven by IT.

Back in the past there was a scepticism on the IT side about RPA mainly due to “low-code approach” and a lot of drag & drop functionalities which was unusual for standard coding. For example screen scraping has been there for many years but only after realizing its potential by business it got more interest and development as a result.

Implementation of RPA in business varies from the implementation in the IT. The difference is in the main focus and starting point. For business, it means alleviating some sort of pain point and they are focusing on making automation work by any price. IT, on the other hand, pays more attention to the technical side with more measured approach – considering the PROS and CONS, infrastructure and whether the Model can work as such.

However, the Operating Model itself does not deviate whether it is implemented in IT or business.

Discussion about the mixed (hybrid) model also took place as this model is getting more attention nowadays, especially after introducing the term “citizen developer”. In this model the CoE only provides the platform (shared knowledge) and consultancy while the end-to-end cycle automation is executed by business units. More details below.

Different CoE models:

Centralized model is basically the “one team for it all”. One CoE has all the functions of the automation and provides the services to the service unit. Here the main advantages include a single point of contact, one methodology, more consistent RPA solutions, centralized support and stronger control. The strongest control can be also listed as a disadvantage as it creates less flexibility, long reaction time. Followed disadvantages include lack of business knowledge and unresponsiveness. This model can be used at the beginning of RPA journey and it’s final part, but its usage is questionable in the scaling part due to limited resources. 

Federated model is located between centralized and decentralized models. There is one team mainly in charge of standards. Other smaller teams are created around the main team (smaller replicates based on geography, industry, function etc.). Main advantages here are increased implementation speed, better economy of scale, higher specific knowledge. However, this comes with increased costs, many times not fully committed people with a lack of RPA skills. This is the best model for scaling.

Decentralized model basically means that everybody does what they want and doesn’t care about what others are doing. Here, every business unit has an RPA initiative for themselves and manages itself. Separate teams also tend to work with different tools. (Subjectively) Decentralized model has quite many disadvantages. It creates a risk of inconsistency in RPA solutions and also the difficulty of scaling. It is possible to use this model in the beginning as “a test ride” but it cannot be scaled.

According to Frank you should start with a centralized or federated model, use the federated model for scaling up and wrap up with either federated model or you can come back to the centralized model.

People tend to have similar wrong beliefs when setting up the operating model. They often do not take it seriously enough and do not give it proper care.

Advantages of automation itself are visible also during the COVID-19. Business continuity (BCP) is almost flawless – when you have automated processes there is zero effect correlated with COVID-19. This can be also applicable to other scenarios such as the revolution in the country, climate issues, etc.

At last, Frank and Eduard discussed how the Operating Model 3.0 could look like. Franks’ assumption is that from RPA itself the focus will be shifted to end-to-end automation life cycle with a variety of tools designed for each stage. People’s involvement in the automation itself should also increase in the future.

DOS AND DON’TS for successful operating model:


  • Do the operating model
  • Create a strategy
  • Position CoE
  • Think about the Model
  • Create a Methodology


  • Don’t start without an operating model
  • Don’t create it and then not follow it
  • Don’t keep it unchanged. Operating model should always be evolved