Learn more about the use case:

Objektkultur Software GmbH

Modern Workplace

27.04.2020

Error reduction in and optimization of work processes

What drove the decision to develop an AI-based use case?
There was a great need to speed up work processes, reduce processing errors and generally increase the efficiency of employees. The overall question was: How can the work results of individual employees be improved? How can added value be created for the employees, e.g. through recommendations for action, and thus make their work easier?
The answer here was AI. The requirements could not be solved with the help of classic, analytical methods or algorithms, because the dynamic processes, work content and methods to be improved change daily. The AI is a permanent observer and constantly analyzes new data at the level of the individual employees. In this way, the AI can train itself in the long term and recognize patterns, interpret them independently, and ultimately draw meaningful conclusions (such as issuing the above-mentioned recommendations for action to employees).

What were your expectations or requirements for the AI application?
One of the main requirements for AI is to recognize patterns in the daily work of individual employees and across employees in a very large amount of data.
In addition, the AI is to identify which employee performs which work/tasks most efficiently. These suggestions will then be communicated as recommendations for action. In the long term, this should accelerate and simplify processes and reduce manpower in development.

Main objectives of AI:

  • Recognition of patterns
  • Increasing the efficiency of employees
  • Acceleration of the work processes
  • Reduction of processing errors

Which partner did you rely on with which technology for the development?
Our solutions are primarily based on Microsoft technologies: We are a Microsoft Gold Partner.

Add-in Core:

  • Azure App Services
  • Azure Cosmos DB/Azure Storage Table
  • Azure Network Security: Secure Network

Add-in AI:

  • Cognitive Services - LUIS: Understanding email content, inferring user intentions.
  • Azure Machine Learning
  • Logic Apps/Function Apps: Pre-processing of email content to make it easier for AI to read (e.g. cutting away salutations, signatures).

The use of AI solutions must also be accompanied by a change in competences. How have you dealt with this challenge in respect of your employees?
We show companies where they can use AI to achieve improvements for employees and thus for the entire company. This process goes hand in hand with active change management.
An important step in the change management process of an AI implementation relates to understanding about the data that is already there and what it means.
Understanding the data that is there every day, but has not been used until now, becomes important through the use of AI. Before, the focus was on understanding about the processes that were taking place; now that takes a back seat. To successfully train an AI, you need the content of the data generated every day. Reason: The AI can recognize patterns, regularities and anomalies in the everyday data that a human might not have recognized before.
This reversal in the understanding of everyday data is an important part of change management and must be transparent for all employees.

In what time frame did you implement the AI solution in your company?
The solution listed here was developed in nine months using an agile approach, introduced as part of a beta pilot, and then rolled out step by step.

Process steps:

  1. Start: Proof of Concept
  2. Minimum Viable Product (MVP): Beta pilot with test persons
  3. Productive roll-out.

A conclusion to your AI solution:
With the help of AI, we were able to create a solution that could not have been developed in the same way with classic tools and analytical methods efficiently or at a reasonable cost.
The advantage is that the AI also recognizes complex changes in the data, develops itself further with the changes and adapts to the changed framework conditions within defined limits. In classical development, the programs would always have to be adapted manually in the event of changes to the framework conditions.
Overall, AI supports flexible and mobile working and brings a company one step closer to modern process automation.

One final piece of advice/tip to other entrepreneurs looking to apply AI:
Before starting an AI project, it is important to consider whether the solution requires the use of AI. After all, AI is not a cure-all for all problems. At the beginning of the AI project, one should start with a task that can be expected to produce a first usable result in less than three months. For this, an external/experienced AI specialist should be brought on board.
During the project, constructive-critical reviews of interim results should be scheduled on a regular basis to prevent project failure.