Provider¶
Zammad offers AI-powered features to support your agents!
To use AI features in Zammad, you must first configure and enable an AI
provider. Such an AI provider processes all your AI requests and is required for
any AI functionality. Add your provider in Zammad’s admin settings
under AI > Provider and make sure to activate the toggle. This setting
requires admin.ai_provider permission to access it.
Note
Zammad’s AI feature is completely optional and requires a configuration before any AI request is made. However, if you don’t want to see the settings at all, read about how to hide it in the system docs.
Settings¶
- Provider
You can choose from different supported providers:
Zammad AI
OpenAI
Ollama
Anthropic
Azure AI
Mistral AI
Custom (OpenAI Compatible)
Depending on which provider you choose, Zammad displays different fields you can configure.
Tip
Keep in mind that your AI provider might require some configuration before you can connect to it. That is the case for Azure AI, anyway. As this configuration is beyond the scope of this documentation, please refer to the documentation of the service you are using.
Note
The AI service you have subscribed to is not available? This is just the first step. We are going to add more providers soon! If you want to support us or need the provider as soon as possible, just let our sales department know!
- Token
Add your API token which you got from your provider here. Don’t confuse it with the tokens a large language model (LLM) is dealing with when processing a request.
- URL
Provide the URL or IP address of your provider. This is required for Ollama and OpenAI compatible providers.
- URL (Completions)
Only for Azure AI as provider. Add the endpoint where your text generation model is available.
- Model
Enter the exact name of the model here. This model has to be available on your provider’s side. Leave it empty (if not mandatory) to use the default which is shown as a placeholder.
- Recognize Image Text (OCR)
Zammad allows you to use optical character recognition (OCR) to extract text from images. This can be useful if you often get error messages or other important information, which is only present in images.
Simply switch the toggle on or off. Be aware that using this feature can cause a high consumption of tokens.
- OCR Model
Specify a model to use for the OCR feature. Leave it empty to use the default model of your provider.
Note: the field is called URL (OCR) for Azure AI and expects a URL where the OCR model is available instead of a model itself.
After providing the information, make sure to click the Save button and
activate AI features.
Excursion¶
In case you want to know more about Zammad AI or Ollama, read on here. Otherwise, head over to Feedback & Logs or check out the AI-powered features and how to use them.
Zammad AI¶
Using Zammad AI as a provider is the easiest way you can get started with Zammad’s AI features. It has some big advantages like:
You don’t have to take care of the setup of an AI system and its configuration.
Hosted in the EU and compliant with GDPR.
No configuration in Zammad needed (for SaaS customers) or only an API key required (for self-hosted customers).
Your requests with sensitive information aren’t used for training.
If something isn’t working as expected, you just have one place where you get support.
- SaaS Customers
Using Zammad AI requires a “V2” plan. Check your subscription settings and consider to switch your plan. If a “V2” plan is active, you can buy AI calls for AI processing. After doing so, Zammad AI is automatically configured as AI provider and you can activate and configure AI features as you like. Check your remaining AI calls from time to time or activate the notification to be alerted when few calls remain.
- Self-Hosted Customers
Please get in touch with our sales department to obtain a Zammad AI API key and to buy AI calls. After receiving your API key, simply add and save it. Then you are ready to activate and configure AI features as you like.
Ollama¶
Ollama is an AI tool that allows you to set up your own AI server on a machine you control. That means your data is not transferred to a third party. Be aware that this only makes sense if you have a powerful GPU in your system!
If you don’t have an Ollama running already, their documentation is a good starting point. You can find the available models in Ollama’s model library and useful information and additional links in the Ollama Github repository.
Feedback & Logs¶
At the top of the settings page, you can switch to the Feedback & Logs tab.
Downloads¶
Your agents can provide feedback like thumbs up or down and leave a comment in case they gave thumbs down. You can download this feedback as well as error logs. Both downloads provide an Excel file (.xlsx) with the requested information.
Recent Logs¶
Zammad shows the last requests and responses in the UI as well. So in case something isn’t working, have a look there. When you open an entry, you can find the request as well as the response from your AI provider.
