<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=150612&amp;fmt=gif">

Is RPA the Same as Artificial Intelligence (AI)?

Posted by Brendan Canny in: robotic process automation, Automation, Machine Learning, Artificial Intelligence, RPA

In my previous blog, I introduced the concept that RPA and our Automation Engine should be seen as a robust “toolkit” used to re imagine the tasks we currently assign to human beings. Artificial Intelligence (AI) and Machine Learning (ML) etc. fit into this concept as simply being additional tools within one arsenal. I’ve heard people refer to RPA processes as being, for want of a better expression, “dumb robots”. RPA sometimes gets this bad rap as, generally speaking, the majority of processes leveraging RPA tools require structured, rules based, process models and the process itself relies on these rules to make predetermined “choices”.

If you combine automation tools with AI / ML you end up with the best of both worlds

AI and ML on the other hand aim to model their “choices” on the human mind, in other words, apply the concepts of intelligence, learning and cognition to decision making, instead of simply following the rules.

Each approach has its own strength, however, and, if you combine automation tools (RPA or otherwise) with AI / ML you end up with the best of both worlds. Structure and time savings from automation, as well as synthetic “thinking” and “learning” around the more “fuzzy” components of any given process. A good case study of this synergy can be seen in commercial lease abstraction automation.

Without AI / ML the, very high level, RPA process would go something like this:

  • A scanned or electronic lease document is saved on a share location or document repository or is emailed to a monitored email account
  • OCR or XML tools are used to “read” the content of the lease
    • This process relies on the text or information to be found in a fairly specific location within the document flow
  • A human user then reviews the content for accuracy and “signs off” on the abstract
    • The human user would need to correct any erroneous data
    • If the error was due to a change in document format then a change to the logic in step 2 would also need to be made to prevent the error reoccurring
  • The digitized content is then mapped and potentially entered into one or many other systems

It is clear that the above process very much relies on the OCR or XML reader to find expected content within a specified location. This means that there will be many leases that will be flagged as problematic within the process, as we all know how complex and ever changing commercial property leases can be.

Enter AI / ML and a new process flow is born:

  • A scanned or electronic lease document is saved on a share location or document repository or is emailed to a monitored email account
  • An AI / ML engine “reads” the contents of the lease applying previously “learnt” information and applying context and natural language to paragraphs
    • The AI engine does not require any content to be in specific places or formats, but rather scans the entire document looking for data points it is expecting to find. The engine does not care if the renewal options are on page 1, or page 100, it just cares that it’s looking for text that appears to reference renewal information
  • A human can perform a quick review of the abstract, as the AI engine would have provided a score of how accurately it thinks it has identified a given data point, meaning that a human reviewing the output can focus on data points that have a low level of confidence
    • Should the AI engine make a mistake, the human user would then “teach” the engine to identify the correct data point – which the AI engine would remember and apply to the next run, thus improving with each lease it processes
  • The RPA tool then takes over and the digitized content is mapped and entered into the required systems

The real value is choosing the best tools to complete the job for each client

These two similar, but critically different, processes highlight the synergy available when combining individual tools (RPA and AI) into a process. Many RPA software platforms are already beginning to build in interfaces into some of the already available AI tools such as IBM’s Watson and Google’s language translation service.

It is for this reason that you will hear me speak passionately about our Automation Engine, rather than just RPA or AI etc. We believe the real value is choosing the best tools to complete the job for each client. The most exciting part is that the number of tools for the toolkit are growing on a daily basis, meaning that more and more processes will soon be able to be handed off to the Rob Sparkes of this world.

The Open Box Real Estate Automation Engine

Find out more