Gladsaxe Municipality selects Applai as their Generative AI partner - KL webinar series on generative AI

KL's webinar series on generative AI includes various sessions focusing on the use and development of AI in municipal contexts. One notable case study is from Gladsaxe Municipality, where they explore the practical application of AI through specialized chatbots that assist employees in their daily tasks.

This initiative is part of a broader effort by the Danish Language Model Consortium, led by IBM, Danish Business Authority, and Alexandra Institute, to develop future Danish language models through public-private collaboration. Read more here. See the full webinar here (case study with Gladsaxe starts at 53 minutes).  

We feel very honored that Gladsaxe has chosen Applai as their partner for generative AI. During the webinar, Thomas Lerche (Digitalization manager at Gladsaxe Municipality) presented his findings from the early stages of the project with Applai. Here are his key findings: 

The problem

Gladsaxe Municipality asked their employess what they were missing from ChatGPT. In other words, what use cases could not be solved using ChatGPT? The employees responded saying things like; "We would like an easy way to get clear guidance for how to use AULA." and "we are 80 caseworkers who need to quickly understand the new 'Childrens law' so we can reduce the time spend searching for information."

The benefits

• Quality in responses
  Specialized AI assistants can deliver higher quality responses that are tailored and trained for specific fields.

• Credibility and transparency
  The assistants offer a higher degree of credibility by citing their sources and making fewer guesses.

• Streamlining workflows
  Specialized AI assistants can support and streamline specific workflows and assist in training new employees. 

The requirements 

• Option to switch language models
  We do not want to be locked into specific technologies or language models that quickly become outdated.

• Easy setup of new assistants
  It should not require external consultant hours or consume many internal resources to train and set up assistants.

• Ensure control and data security
  We want to store training data and logging data ourselves and use "our own" language model.

Three step implementation

• Train
  Load data from local files, websites, or the intranet and use them to train a specialized AI assistant.

• Build
  Choose a language model, define rules and the assistant’s role, and give it a user-friendly appearance and behavior.

• Share
  Provide controlled access to the assistant via the municipality’s own access management system (AD FS).

Test examples

Solution overview 

• Data is processed exclusively by the municipality itself and, if applicable, the LLM host (e.g., Azure).

• The web interface for building chatbots is continuously maintained with new features and always supports the latest AI models.

• A complete secure solution that integrates with the municipality’s existing security setup.

Product demo

Status and process

1. Technical Implementation
   Three specialized AI assistants are now operational and ready for implementation

2.  

   Assistants in Operation
   The three assistants will be tested until the summer holidays and experiences will be gathered

3.  

   Introduction and Training
   The three departments are introduced to the new assistants and receive training in their use

4.  

   Implementation in a Broader Perspective
   Strategic discussion - where do we need assistants, what are the consequences, etc.?

Easy to Get Started

• The solution can be quickly deployed in the municipality's IT environment.
• The municipality is responsible for the on-premises backend environment (following guidance from Applai).
• Applai installs, maintains, and updates the web interface, etc.

Price and Setup
With the municipality's own LLM instance on Azure and self-hosting of data.

In addition to the license for Applai, there are costs for LLMs, which are settled directly with Azure.

Contact for more information:

Applai 
Applai.dk
Victor Skytte, Co-founder 
Victor@applai.dk 
+45 28 76 22 98

Thank you for reading and please share this blog post.

🫶 
Mathias