The meeting was held on 11/09/2026
Thank you for another good meeting in the Promte User Club. On 11 September, we brought together users, partners and organisations considering Promte for product updates, demos and a shared discussion of everyday needs.
This time, we focused on building apps directly in Promte, the opportunities offered by Danish and European language models, and better oversight of how organisations use AI.
You can watch the recording from the user club meeting above for the full walkthrough, demos and questions from participants. The recording is in Danish.
We welcomed Dana, who has joined as an AI solutions consultant, and Harun, who has joined as a developer. Dana will take part in workshops and help make solutions easy to use and relevant to organisations. Harun works on developing and improving the Promte application.
One of the major updates was Promte Kodning, our coding module. Users can describe the solution they need to a coding agent and follow the work on the app directly in Promte. This brings development together in one place, where users can also continue working with files and code.
Apps can have both their own user interface and functions that run in the background. They can connect to APIs and use the features and data sources the organisation already has access to in Promte. Examples include uploaded files or searches across data from Microsoft integrations and SharePoint.
We also showed how governance is part of the process from development to publication. An app can be submitted for administrator review, providing an overview of its purpose, data processing, use of AI and relevant policies. The review is based on the code that has actually been written.
Administrators can approve or reject access to features and external services, restrict permitted code packages and control which data may be sent to language models. Guardrails are also integrated into the app environment, allowing specific data types, such as company registration numbers, phone numbers and addresses, to be blocked or logged. The activity log provides insight into events and checks, and consumption can be limited at organisation, department and user level.
We were joined by Corti, which presented opportunities to use language models hosted in Europe and, in the future, on Gefion in Ballerup. The prospect is to enable data processing in Denmark across Promte Kodning, assistants, chat and meeting minutes.
Corti discussed testing models for coding and agent tasks, as well as its work on security, encryption and quality control, drawing on its experience in healthcare.
We also covered the arrangements for adopting the models. For organisations whose Promte installation is hosted by us, using Corti will require an addition to the data processing agreement. Organisations hosting Promte themselves will be able to enter into a direct agreement and connect the models to their installation.
We also reviewed a series of updates designed to make Promte easier to use and administer:
The analysis agent has also been improved. It can plan its information search and review larger amounts of data before responding. This is useful, for example, when examining agendas and minutes over an extended period. An existing assistant can be converted into an analysis agent and continue working with the same data sources.
Promte's documentation has a new setup with guides for both employees and administrators. We also demonstrated an upcoming help feature in the application that suggests relevant guides based on the page the user is viewing.
The guides include explanations, examples and screenshots, and the documentation is set up to follow product changes automatically. The help feature was presented as being close to launch.
Promte Styring, our governance module, gives administrators a consolidated view of activity, workspaces, model usage and token consumption. The module also brings together data protection settings. Conversations in selected workspaces can be made private, and user identities can be anonymised if the organisation wants insight into the content without seeing who wrote it.
Organisations can also choose which language models are available generally or within specific workspaces. This makes it possible to try a new model in a limited environment before making it available to more employees.
Participants requested further options for risk assessment, reporting and sharing data with management. We will take these suggestions into the continued development of Promte Styring.
We also shared a look at Promte Desktop, which is under development. The desktop application is intended to let analysis agents work with local files, summarise multiple documents and create new content. We would like to hear from interested organisations about potential uses, file access and requirements from their IT teams. Read more about Promte Desktop.
We also presented a hackathon programme where several organisations can come together for an in-person day introducing vibe coding and practical work in Promte Kodning. The programme includes two months of access to the module and credits for the models. After one month, we follow up so participants can present their solutions and receive feedback.
Read more and register for the hackathon programme.
We will also hold a webinar on Promte Kodning and vibe coding on 29 September, 9–10 a.m. Danish time.
Read more and register for the webinar on Promte Kodning and vibe coding.
Thank you to everyone who participated, asked questions and shared experiences and ideas. The user club helps us develop Promte in close collaboration with the organisations using the platform in practice.
If you have ideas for apps, requests for Promte Styring or would like to learn more about Promte Desktop and the hackathon programme, please get in touch.
