S5/S7 AI for Windows:Copilot access to AI

Copilot in S5/S7 AI for Windows® can use several AI services. It is not limited to ChatGPT or OpenAI.

Select the required connection in the Provider field. Credentials, server addresses and models are stored separately for each provider.

Available providers

Provider	Connection	Required information	Typical use
IBHsoftec Server	Copilot server provided by IBHsoftec	Copilot license or activation of the displayed PC fingerprint	Simple central access without a personal AI API key
Company Server	Copilot server operated by your company	Server URL and activation of the PC fingerprint	Centrally managed company access
OpenAI Direct	Direct connection to the OpenAI API	OpenAI API key and model	Use of your own OpenAI account
Google Gemini	Direct connection to the Google Gemini API	Gemini API key and model	Use of your own Google AI account
General LLM	OpenAI-compatible API	Server URL, API key and model	Local or private services such as Ollama, Open WebUI, vLLM or LiteLLM
Claude Direct	Direct connection to the Anthropic API	Anthropic API key and Claude model	Use of your own Anthropic account
Microsoft Foundry	Direct connection to Microsoft Foundry	Foundry endpoint, API key and Azure deployment name	Models deployed in Microsoft Azure

IBHsoftec Server

Use this provider when Copilot access is to be routed through IBHsoftec.

1. Open Copilot.
2. Select IBHsoftec Server.
3. Purchased software includes four weeks of free full access.
4. Limited access for a demo version can be requested from IBHsoftec.
5. If activation is required, send the displayed Fingerprint to IBHsoftec.
6. Use Copilot after activation.

The server URL is preset in the program. No personal OpenAI or Anthropic API key is required.

Important:

• Activation applies to the displayed PC fingerprint.
• The PC requires Internet access.
• After the four free weeks, continued access costs 49 euros per month.

If the server publishes a model list, Copilot uses that list. Otherwise, Server default is displayed. In this case, Copilot sends no model name and the server uses its configured default model.

Company Server

Use this provider when your company operates its own Copilot server.

1. Open Copilot and select Company Server.
2. Enter the URL supplied by the administrator.
3. Send the displayed fingerprint to the administrator.
4. Use Copilot after the PC has been activated on the server.

The user does not need a personal AI API key. The company centrally controls models, access permissions and usage.

OpenAI Direct

Use this provider with your own OpenAI API access.

1. Create an API key for the required OpenAI project.
2. Select OpenAI Direct.
3. Enter the API key.
4. Select a model and send the request.

The model list is retrieved automatically from OpenAI. Costs, quotas and model permissions depend on the OpenAI project used.

API keys must not be shared or included in projects, screenshots or documentation. The voice dialog is currently supported only with OpenAI Direct.

Google Gemini

Use this provider with your own Google Gemini API access.

1. Create a Gemini API key in Google AI Studio.
2. Select Google Gemini.
3. Enter the key in the Gemini key field, or provide it through

 the GEMINI_API_KEY environment variable.

1. Select a Gemini model and send the request.

Copilot retrieves the available Gemini models automatically. The voice dialog is not available with Google Gemini.

General LLM

General LLM connects Copilot to an OpenAI-compatible API. It can be used with services such as Ollama, Open WebUI, vLLM and LiteLLM.

1. Make the service available to the PC over the network.
2. Select General LLM.
3. Enter the base address in Server URL.
4. Enter the API key intended for that server.
5. Select a model and send the request.

Copilot retrieves the available models automatically. For Ollama and Open WebUI, model details and context size are also read when available.

The reduced initial project context for smaller local models applies only to General LLM, not to other providers.

Claude Direct

Use this provider with your own Anthropic API access.

1. Select Claude Direct.
2. Enter the Anthropic API key.
3. Select or enter the required Claude model.
4. Send the request.

Claude Direct supports text requests and supported image, PDF, text and source code attachments. The voice dialog is not available with Claude Direct.

Microsoft Foundry

Use this provider for a model deployed in Microsoft Foundry or Azure.

1. Select Microsoft Foundry.
2. Enter the API key and Foundry endpoint.
3. Enter the exact Azure deployment name in the model field.
4. Send the request.

The deployment name is not necessarily the public model name. The message The API deployment for this resource does not exist usually means that the entered name does not match a deployment in the selected resource.

Inference API keys cannot reliably list the available Azure deployments. Therefore, the deployment name is entered manually.

Model lists and context size

Models are retrieved automatically for:

• OpenAI Direct;
• Google Gemini;
• General LLM;
• Company Server;
• IBHsoftec Server.

The list is updated when Copilot is opened, when the provider changes and when the model list is opened.

If the provider reports a context size, Copilot displays the maximum number of tokens. Not every service exposes this information through its model API.

Security and project data

• API keys and Company Server fingerprints are masked in the user interface.
• Credentials are stored separately for each provider and protected in the Windows user profile.
• Provider settings are retained when switching to another provider.

Depending on the request, Copilot may send project information, diagnostic data, requested live values or selected attachments to the chosen AI service. For confidential projects, use only a provider approved by your organization.

Further documentation

• Copilot user guide
• Complete diagnostic example
• Set up a Company Server
• What's new?