Blog Single

18 Aug

What is intelligent automation?

We have already tried in the last articles to present the basics of robotics as simple as possible and to show which things lead to more transparency and acceptance. No matter which technology, which provider or in which way you use automation, it always requires an adaptation in your company (often described as change management or organizational development) and your processes.

A company that hasn’t had much to do with automation till now must above all make and keep its goals transparent (have a look at Help, the robot is coming! or Robotics Reporting: Create transparency!).

Their processes were never designed for automation and therefore, need to be adapted, both in terms of objectives and scope. Nevertheless, there are always surprises. Use automation as an opportunity to first think about your process results before you go further and define the “happy” path to the goal. Often the first approaches to solutions are too short and not thought through till the actual final result.

In addition to these fundamental topics, we are also repeatedly confronted with the questions “How can I implement intelligent automation? What do I have to consider and is the “classic” robotics not already obsolete?

First of all, no, the “classic” robotics is not replaced by it. On the contrary, it can even be a useful addition to an intelligent solution.

What has to be taken into account for intelligent automation?

Even with intelligent automation, there is no way around the basic topics such as process clarity and definition, integration into your company and everyday work with robotics.

We ourselves recommend not to start directly with intelligent automation, because you have to trust a system very much to give it such complex decision-making powers. In the end, these solutions will also make decisions for you, except that the processes and the path to the decisions are more complex than with “classic” robotics. We have already developed solutions that work even with little experience in robotics or automation topics, but it depends even more on your organization how effectively such solutions work. In any case, the technology is almost never the limiting factor.

So how can intelligent automation be implemented in practice?

If you have a specific need, for example in the categorization of mails for the allocation of teams, posting of incoming invoices, or the automatic creation of lead lists for sales based on online information, then you have already taken your first important step. Because without a very specific use case with a clear goal, intelligent automation is difficult. The more “intelligent” the automation is to be, the more precisely the framework conditions in which decisions are made must be determined. There is currently no (market-ready) software that “learns” its processes and then intelligently automates them.

Once the need has been clarified, there are two possible approaches to their intelligent automation:

  • There is already a special solution (software or platform) for your specific application
  • You (or an external partner) develop the required “intelligent model” yourself (in the sense of data science by means of machine learning or artificial intelligence). An automation solution can then be linked to this model. This can be done either with a specially programmed solution or with robotics, whereby the latter is based on the mutual support of technologies or solution approaches.

What exactly do I get from intelligent automation?

For example, they can use learning effects that were previously not available. For example, you can develop, operate, and further refine a model for posting incoming invoices so that the incoming invoices are ultimately posted optimally by a system. How the system learns how to set tax codes, G/L accounts, vendors, and so on, can be mapped using an intelligent model and can be developed to the point where, at the end of the process, all that remains is to confirm whether the goods or services have actually been delivered in the quantity and quality billed.

Another use case is the categorization of mails. Here models for a certain text comprehension can be developed from which categories can be derived. For example, you can forward some mails to a team for processing, while other mails can be processed in advance and transferred to an automation solution. For example: mails to customer service with a complaint should be forwarded to clerks, mails with registrations should be checked for completeness and then be processed directly.

So you have very special use cases, but in such use cases there are usually also very high transaction volumes (frequent execution, long process durations, …). This allows intelligent automation, but the models must be maintained and refined further. As already mentioned at the beginning, after a transition phase, a (so optimized) system will at some point have to be trusted to the extent that the intelligent automation can actually work.

We hope to have given you again a first general overview of the implementation possibilities of robotics and intelligent automation and what they offer you. We would be pleased to support you in the implementation of a pilot project, the selection of suitable software, accompany you in the selection of suitable process candidates for automation, up to setting up an effective reporting system.

As always, if you have any questions about automation, its implementation, integration into your process landscape, or the selection of an RPA provider, please contact and we will be happy to help you.

Your WorkAnizer Team

Related Posts