3D Construction Printing (3DCP) is quickly gaining traction in the construction industry due to its promises within logistics, design and material saving. Another significant potential is the need for few but qualified personnel, that will be able to 3D-print houses, installations and infrastructure. Quality assurance and quality control (QA/QC) are needed to ensure that critical structures are in fact of the quality they need to be. This can be especially important for e.g. bridges and foundations. However, the time spent on QA/QC will increase dramatically, as a percentage of total production time, when the personnel involvement in 3DCP is far less than conventional processes. This may weaken some of the promises of 3DCP. In this project, we develop workflows involving Vision-Language Models (VLMs) to be used for QA/QC of the 3DCP process. The model is being fed recordings of the printing process together with key sensor data, such as pressure, temperature, and pump speed. Based on this data, the model concisely and objectively describes the printing process, highlighting any abnormalities that may occur – e.g., a significant delay between an action being called and the action occurring. The developed system serves to automate the reporting of the production process, as well as to troubleshoot and assess the severity of abnormalities. A prototype has been tested on data from the 3DCP system at Mechatronics Innovation Lab. This presentation will discuss details of the workflow involving VLMs, how this may change the process of 3DCP, and other findings.