AI / LLM configuration
Turn AI features on, choose between the managed NetLock AI and your own OpenAI-compatible provider, and control where AI shows up in the product.
AI / LLM configuration
The AI / LLM Settings page at /settings/ai is where a deployment operator turns AI on, decides where inference happens, and controls where AI appears in the product.
Two model sources are available, and the choice between them is made on this page: the managed NetLock AI (hosted on IONOS AI Hub), or an OpenAI-compatible provider you run or subscribe to yourself. The second option uses the OpenAI chat-completions API, so any provider that offers it (OpenAI itself, Anthropic via its OpenAI-compatible endpoint, Ollama, LM Studio, LocalAI and others) works with the same fields.

A.11.1 The master toggle
Enable AI Features sits at the top of the page. It decides whether any AI feature in the product is active.
- When off, all AI features in NetLock RMM are disabled. The AI Chat page is unavailable, AI buttons inside Scripts and Remote Shell are hidden, the
Analyze with AIbutton on Event Details is gone, the AI Assistants inside the Add / Edit Sensor and Add / Edit Job dialogs are hidden, and the end-user chat in the agent tray is refused by the server. Everything else in the product keeps working. - When on, the per-feature toggles below decide which AI features are active (see A.11.4).
Save stores the change. Switching the master toggle does not change the model selection or the provider configuration below; when you turn AI off and back on later, the settings you saved last apply again.
Tip: If you are not sure whether AI is the right fit for your deployment, turn the master toggle on, decide on a model source, and leave every per-feature toggle at its default. Individual operators can then explore AI Chat (Chapter 13) before you roll out the AI buttons on the other pages.
A.11.2 Model selection
The Model Selection block is a two-option radio group directly below the master toggle. It decides where every AI request from this deployment is processed, which makes it the most important setting for a compliance review.
Use NetLock AI (managed)
The default for a new deployment. Requests do not go to a provider of yours: the Web Console (and, for the end-user tray chat, the server) sends them to NetLock, which forwards them to its AI provider, the IONOS AI Hub, and streams the answer back along the same path. Authentication is the deployment's Members Portal API key (the same key the Licensing page holds); no provider credentials of your own are involved, and the provider fields described in A.11.3 are hidden because they do not apply.
In this mode the managed service sets the request parameters. The model is the one the managed service runs and cannot be chosen here; the sampling temperature is always 0; and the completion length is capped at 4 096 tokens, or at the remaining token budget of the deployment if that is lower.
Consumption counts against the token budget included with your membership. The page shows it under NetLock AI Usage as a progress bar with <used> / <max> tokens (<remaining> remaining). The same figure appears on the NetLock AI card under Settings → Licensing (A.1). If the membership does not include managed AI, the block reads "NetLock AI is not enabled for your membership." and no request can be sent in this mode.
Use my own model
Every request goes directly from the Web Console (and, for the end-user tray chat, from the server) to the provider you configure in A.11.3. Nothing is sent to NetLock, and NetLock cannot see the content. Model, token limit and temperature are taken from your fields and sent unchanged.
This is the only mode that keeps AI content inside your own infrastructure. If your compliance rules require that prompts, script bodies, event data, shell output, or ticket content never leave your network, select Use my own model and point it at a model you host yourself.
Note: Switching between the two modes does not clear anything. The provider fields keep their saved values while the managed mode is selected, and the managed mode is available again after you switch away from your own model.
A.11.3 Provider configuration
These fields appear when Use my own model is selected, and every AI call the deployment makes in that mode uses them.
Provider Name: a free-text label for your own overview. It is not rendered on any other page and is not sent to the provider. To confirm which endpoint is actually in use, read the active-endpoint line on this page rather than this field.API Base URL: the root URL the request path is appended to. NetLock RMM appends/chat/completionsand sends the result exactly as it stands; it never adds, removes, or corrects a path segment. Enter the OpenAI-compatible root including the/v1segment where the provider uses one; for Ollama that ishttp://<host>:11434/v1, nothttp://<host>:11434/. The provider templates below fill this in correctly; if you type the URL by hand, check it yourself. The helper text on the field readse.g., https://api.openai.com/v1.API Key: the secret sent in theAuthorization: Bearerheader. Password-masked input. Leave it empty for a local provider that does not require authentication: with an empty field noAuthorizationheader is sent at all, which is what Ollama, LM Studio, and comparable runtimes expect.Model: the model identifier the provider expects in the chat-completions call. The helper text readse.g., gpt-4o, claude-sonnet-4-20250514, llama3. If you configure an identifier the provider does not recognise,Test Connectionshows the provider's error message.Max Tokens: the maximum response length. Numeric input, range256to128000. The value should match what your chosen model and provider accept. A local model with a 4 000-token context window will not honour a128000setting. The default produced by the provider templates is4096.Send temperature: toggle. When on, aTemperatureslider appears with range0.0to2.0in steps of0.1and the value is sent with every request. Lower values produce more predictable output (at0.0the model always picks the most likely next word); higher values produce more varied output.0.7is a reasonable default for script analysis and event commentary, which is what the provider templates set. Turn the toggle off for models that reject requests that carry a temperature; the parameter is then not sent at all.
Provider templates
Four one-click templates populate the fields with sensible defaults. Clicking a template overwrites Provider Name, API Base URL, and Model, resets Max Tokens to 4096 and Temperature to 0.7 with Send temperature on, and leaves API Key as it is. A local runtime does not need a key.
| Template | API Base URL | Model |
|---|---|---|
OpenAI | https://api.openai.com/v1 | gpt-4o |
Anthropic | https://api.anthropic.com/v1 | claude-sonnet-4-20250514 |
Ollama | http://localhost:11434/v1 | llama3 |
LM Studio | http://localhost:1234/v1 | local-model |
For a provider that is not listed above, for example LocalAI or a self-hosted vLLM, enter its API Base URL and Model manually. There is no "Custom" template; the four templates only prefill the fields.
Note: For Ollama and LM Studio, the template URL uses
localhost. That address points to the host running the Console, not to your workstation and not to the device running the agent. If you run a local model on a different machine, replacelocalhostwith the reachable hostname or IP. If you run the Console in a container, make sure the container can reach the host's loopback if that is where the model lives.
Test Connection
The Test Connection button sends a minimal probe with the current field values. While the call runs, a spinner labelled Testing... is shown. The result appears below the button, either as a success alert or as an error alert that quotes the provider's error text.
Use Test Connection every time you change API Base URL, API Key, or Model. Save only after the test succeeds. An untested configuration can make every AI action fail.
A.11.4 Per-feature toggles
Ten toggles decide which AI features are active in the rest of the product. All of them are on by default. They can only restrict what the master toggle allows: a feature toggle that is on has no effect while Enable AI Features is off.
| Toggle | Where the surface appears |
|---|---|
Script Analysis | The AI Assistant in the Add / Edit Script dialogs. See Chapter 8.1. |
Remote Shell | The AI actions inside the Remote Shell dialog and inside Bulk Remote Shell. See Chapter 3.5. |
Event Analysis | The Analyze with AI button on the Event Details dialog, both on the device Events tab and on the Events page. See Chapter 12.2. |
Sensor & Job Creation | The AI Assistants inside the Add / Edit Sensor dialog and the Add Job dialog. See Chapter 8.2 and Chapter 8.3. |
Event Log Analysis | The AI buttons in the Remote Event Log viewer and on an event-log entry's details. See Chapter 3.3. |
Ticket Analysis | The AI actions in the ticket detail dialog. See Chapter 10. |
Audit Analysis | The AI analyze action on the Audit page. See Chapter 12.3. |
Automations Assistant | The AI Assistant on the Automations page and in the Add / Edit Automation dialogs. See Chapter 5. |
AI Chat | The operator AI Chat page, the slide-in chat panel, and the chat button in the top bar. See Chapter 13. |
End-user Chat | The end-user AI chat in the agent's tray icon. This one is enforced on the server: with the toggle off the tray's requests are refused and the menu entry disappears on the tray's next sign-in. The tray entry additionally depends on the AI chat switch in the tray-icon policy, which stays an independent second switch. See Chapter 13.6. |
Turning a toggle off hides the AI element wherever it appears; the feature it belongs to keeps working. With Script Analysis off, the Scripts editor works normally, only without the AI Assistant.
The toggles are deployment-wide, not per user. Use them to decide which of your data may reach the model at all. Per-user permissions cannot do that, because they describe what a person may do, not what the model sees.
A common rollout is to leave AI Chat on for all operators (it is the cheapest feature to evaluate), enable Script Analysis and Event Analysis once the team is comfortable, and keep Ticket Analysis and End-user Chat off until the content these features send has been reviewed.
A.11.5 Documentation grounding
The Web Console contains the product documentation of its own release (the chapters of this manual as of the build) and uses it to answer questions about NetLock RMM. When an operator asks the AI Chat how a page, a setting, a policy or an agent behaves, the console picks the documentation sections that match the question and sends them to the model together with the question. The model is instructed to answer from those sections, to quote page and setting names as written there, and to say when the documentation does not cover a question instead of guessing. The answer ends with a Sources line that links the documentation pages the model cited.
Two fields under Product documentation control this:
Use the product documentation to answer questions about NetLock RMM: toggle, default on. When off, the chat sends the question without documentation sections and the model answers from its own knowledge.Documentation excerpt budget (characters): numeric, range2000to32000, default8000. The maximum number of documentation characters the console adds to one request. A section that does not fit the remaining budget is left out.
A read-only line below the fields names the documentation snapshot the console carries: the product version, the number of articles and sections, and the revision and date of the documentation it was built from. When the line reports that no snapshot was found, the installation was built without the documentation and answers come from the model alone.
Which chats use it:
- The
AI Chatpage and the slide-in chat panel opened without a context always run the lookup and send up to the configured budget. - The context chats (the assistants in the Script, Remote Shell, Event, Event Log, Audit, Sensor, Job, Automation and Report Builder dialogs) run the lookup as well, but only add documentation when the question clearly matches a documented topic, and with at most 3,000 characters, because those chats already carry the script, log or schema they were opened for.
- The ticket AI actions and the end-user chat in the agent tray never add documentation.
The screenshots of this manual ship with the console as well. When a documentation section handed to the model carries a screenshot of the page or dialog in question, the answer can show it below the step it illustrates. The image is served by the Web Console itself from the bundled copy and opens in full size in a new tab; nothing is loaded from the documentation site or any other external source. Images from anywhere else are never rendered in an answer, only their description text appears.
The lookup runs inside the Web Console. Nothing is sent anywhere except the request to the configured model, which carries the matching documentation sections alongside the question. On the managed NetLock AI a grounded question adds roughly 2,000 tokens, which count against the membership budget like any other prompt text; on your own model it adds the same amount to the prompt the provider sees.
Note: Small local models with a short context window cannot hold the default budget together with the conversation. Ollama, for example, truncates a prompt that exceeds the model's context length from the start, which drops the instructions and the documentation sections before the question. The answer is then not based on the documentation, although everything is configured. Either lower the budget to
2000or3000, or raise the model's context window (for OllamaOLLAMA_CONTEXT_LENGTH=16384ornum_ctxin a Modelfile) and keepMax Tokenswell below it. A temperature of0.2to0.4gives the most faithful answers on your own model.
A.11.6 Chat history cleanup
AI conversations started from the AI Chat page (see Chapter 13) are stored in the database. Two fields control how long they are retained.
Auto-cleanup chat history: toggle. When on, conversations older than the retention window are deleted on a rolling schedule. When off, conversations are kept until manually deleted from the AI Chat page.Keep chat history for X days: numeric, range1to9999. Disabled when the auto-cleanup toggle is off.
Turning the toggle off stops the cleanup at once. Turning it on resumes it with the next scheduled run. Clearing the retention window does not restore conversations that were already deleted; deleted conversations cannot be recovered.
Warning: Chat history can contain sensitive information operators pasted into the chat (credentials, internal hostnames, PII). If your compliance rules require it, set a short retention window (7 or 14 days) from day one. Longer windows make sense if you use older conversations for training or coaching.
A.11.7 What to verify after changes
After changing settings on this page, check the following:
- Read the active-endpoint line on the page. It names the resolved mode and the target URL as saved, and is the reliable answer to where the deployment sends its prompts.
- On
Use my own model, runTest Connection. Only a success alert confirms that the provider is reachable and that URL, key and model work together. OnUse NetLock AI (managed), check thatNetLock AI Usageshows a budget rather than the "not enabled for your membership" notice. - Open AI Chat (Chapter 13) and ask a one-line question. Confirm a reply arrives within a sensible time for the model source you chose.
- Open a page with an AI feature whose toggle you enabled (the Scripts editor for
Script Analysis, Event Details forEvent Analysis) and confirm the AI button appears and produces output. - With
Product documentationon, ask AI Chat a question about the product ("Which permission is needed to delete a policy?"). The answer should carry citations such as[1]and end with aSourcesline; if it does not, see A.11.5 and the troubleshooting entry AI answers are not grounded.
If step 2 passes but step 3 fails, the endpoint is reachable but the Chat-specific prompt is not being answered; on your own model, check Max Tokens against the model's context window. If step 2 passes but step 4 fails, the per-feature toggle is off or the operator's role lacks the relevant feature-area permission.
Permissions
settings_enabled: access to theSettingsgroup.settings_ai_llm_enabled: access to theAI / LLM Settingspage.
Related chapters
- Chapter 13 — AI Chat — the primary AI surface for operators, and the end-user tray chat.
- Chapter 8.1 — Scripts — AI Assistant in the Scripts editor.
- Chapter 8.2 — Jobs and 8.3 — Sensors — AI Assistant in Add / Edit dialogs.
- Chapter 3.3 — Device details view — AI buttons in the Remote Event Log viewer.
- Chapter 3.5 — Remote control — AI in remote shell.
- Chapter 12 — Events & Audit —
Analyze with AIon Event Details andAI analyzeon Audit. - A.1 — System overview and licensing — the Members Portal API key and the managed-AI token budget.
- X.5 — Integration endpoints — the full list of external endpoints NetLock RMM reaches out to, including both AI paths.