In my previous blog, I talked about the next generation of intelligent automation applications that will be created using AI, Generative AI, RPA, and low-code user interfaces. There has been much discussion lately about what’s next for RPA and how it can work with AI. In April, Google demoed next generation AI Agents, which use a blend of true AI, Generative AI, and RPA functionality to interact with applications. These agents work on top of the applications to ensure a productive desktop experience. I believe that this is the next-generation of desktop automation, taking it a step further and thinking about using RPA to interact with an application on the front and back end and provide decisioning through orchestrated AI, all as part of a seamless simple user interface. One can imagine a very powerful next-generation application. This application will be extremely agile as it will need to change as the human workflow evolves. These will not be static apps but will morph and change to meet the needs of the workflow. In addition, AI will need maintenance and care as it’s completely dependent on high quality data. Models will need to be maintained and updated. Also, RPA will need to be supported as it will play a key role in the Orchestration and integration of the AI Agent.
So where does this lead us? There is a term that references the switch from “Software as a Service” to “Service as Software.” This describes what are referred to as intelligent Automation Managed Service companies, similar to my company, CampTek Software. The necessity of providing end-to-end support for the AI Agents will be imperative. Since these applications are somewhat specific to the end user and their role in the organization there will need to be proper process analysis done up front to ensure the effectiveness of the AI Agent. Part of the scoping process will include the design of the “to be” solution. Scoping will be a more extensive phase as it will involve what type of data is needed for AI, what decisioning is needed beyond typical rules-based logic, and also what the front and backend architecture involves (which applications will feed AI Agent and vice versa). After it has been scoped it will be developed using reusable components as a foundation, so the work resides more in the configuration with model building and User Interface design. Once the AI Agent goes into production, there will need to be ample support for hosting, data harvesting, and application changes (both with the existing desktop and functionality of the AI Agent), not to mention system health and security analysis. The typical Software as a Service approach of buying a finished product that needs minimal support is not possible any longer with these types of incredibly intelligent AI Agents. Service as Software is the model that will prevail as we usher into a new intelligent automation era.
Written by: Peter Camp, CTO & Founder
