An Appian application is comprised of many different processes. They are all designed with the intention to improve the overall process. However, after an application release, it is difficult to know how the overall process is really performing, if the goals are met or what impact a new release has.The ProcessMiner offers a holistic view on your business processes and how they really run.
Artificial Intelligence can be integrated very easily with an Appian solution. However, the direct benefits might not be immediately salient. The ProcessMiner allows to apply AI to applications that currently don't use AI. By this, potential for further automation can be discovered.
What it does
Users can specify their end-to-end processes by selecting Appian process models. Furthermore, they can define all relevant process variables. Google AI is used to analyse text (e.g. sentiment) and images (e.g. labels). The ProcessMiner will store all process instances of the defined process. The user can run an analysis to discover how the processes are really running. For this, the different process variations, performance metrics, and variable values/distributions will be displayed. In order to improve the process, tasks can be created for the process stakeholders. Also, the user can create different analyses in order to compare the performance over time, for instance if a new release needs to be evaluated.
How I built it
It is an Appian application that integrates with Google's Natural Language and Vision API. The data is analysed within Appian's MySQL cloud database.
Challenges I ran into
Tracking an end-to-end process across different process models was a very demanding task as they are only linked by common process variables.
Accomplishments that I'm proud of
When it all came together and I could see the different process variations with the aggregated AI data.
What's next for Appian ProcessMiner
- Working with archived process models
- More analyses dashboards to offer a broader view on the end-to-end process. Here are countless opportunities.
- Using the data to predict the outcome and execution time of processes
- Analysis of all process nodes and not only the tasks
- Features to compare different analyses automatically/guided