Top Artificial Intelligence Companies 2023Top AI Companies 2023Top Web Developers 2023Top Web Developers 2023ZF Group is a global technology company that supplies systems for passenger cars, commercial vehicles, and industrial technology. The ZF’s products include driveline and chassis technology for cars and commercial vehicles, along with specialist plant equipment such as construction equipment. The company has 168 production locations in 32 countries with approximately 165,000 employees.
The project consisted of 2 main phases: the ideation and the PoC implementation, followed by the pilot run in one of the plants.
In the first phase of the project, we ran workshops where we analyzed several use cases for generative AI implementation, evaluated them, and prepared a report with a summary of our findings and recommendations regarding the next steps and tech stack.
As the client decided to proceed, we implemented a PoC version of a virtual maintenance assistant to help their employees identify root causes and solutions for maintenance challenges in manufacturing facilities across Europe. The assistant streamlines internal knowledge access, improves data accessibility, and — as a consequence — reduces downtime. Additionally, every answer includes a link to the original information to limit the black box effect, reduce the hallucinations, and increase trust.
First, we needed to analyze several use cases and help the client choose the one to proceed with.
Then, we developed a proof of concept to prove the technical feasibility of the designed solution.
The model needed to search for answers across the databases from various plants based in different European locations.
The topic of artificial intelligence adoption was already on the ZF’s agenda. They had several ideas in their minds. Our first task was to help them analyze different use cases and choose the one that would create business value while being easy to implement. We wanted it to be a quick and cheap experiment that would prove the technical feasibility of the design solution and maximize the chances of success.
Before committing to a long-term and expensive project, we wanted to validate the technical feasibility of using AI for a certain use case. In other words, we wanted to have proof that our hypotheses were correct and that AI can generate business value.
First, we needed to agree on where to start. During the AI Exploratory Workshop, we analyzed several use cases and assessed them by different factors such as:
Finally, we recommended the client to continue with building a PoC Maintenance Assistant as it was the fastest, easiest, and cheapest way to check if generative AI could create business value.
The challenge we attempted to solve was related to production downtimes. In case of any machine malfunctions, it was necessary to call the service (often an external one) and wait until the problem was resolved. Sometimes, it took hours — even if the problem wasn’t very complicated and could have been easily fixed without calling a service.
The AI Maintenance Assistant was created to assist factory maintenance teams with efficiently diagnosing equipment failures. Powered by GPT and Cognitive Search based on the Azure cloud, it was connected to the internal database containing information about past failures and downtimes. Ultimately, after the Pilot phase, it will be integrated with multiple data sources — data acquired from different plants — regardless of their language and data structure.
The collected data was indexed and organized in a searchable database, facilitating quick and efficient retrieval and analysis of historical maintenance records. When asked a question, the assistant extracted information from the user inquiry and searched through the database to help users securely access internal documentation and get the insights they needed.
Using structured and unstructured text data, such as maintenance notes, the assistant facilitates the rapid identification of the underlying causes behind equipment malfunctions. It provides ZF staff with relevant, summarized information, answers their queries, and offers suggestions for resolving common production and maintenance issues.
The PoC gathered positive feedback from the stakeholders and was sent to one of the plants so it could be tested by local managers. Once we gather users’ feedback, it will be possible to expand the solution, add necessary features, and plan the company-wide adoption.
In the future, we plan to expand the feature’s capabilities, addressing technical challenges and refining the user experience based on the collected feedback. Additional challenges will certainly arise as new data sources — in different languages and with different data structures — will be added.
Top Artificial Intelligence Companies 2023Top AI Companies 2023Top Web Developers 2023Top Web Developers 2023