Get started with the AI Center in rConfig V8

This page takes you from a fresh rConfig install to a working AI answer about one of your device configs, in about fifteen minutes. Each step links to a page with the full detail. Come back here if you get lost.

The AI Center is available from V8.3.0. The terms acceptance step was added in V8.3.3.

Before you start

You need:

• Shell access to the rConfig server, to set one environment variable.
• An account with the aicenter.update permission, such as an Admin. Adding a provider and turning capabilities on need it.
• Something to answer the questions. That’s either an API key for a cloud provider (Anthropic, OpenAI, Google Gemini, or Azure OpenAI), or a model server you run yourself (Ollama, or anything that speaks the OpenAI API).
• At least one device with a downloaded config, so you have something to ask about.

Keeping config data on your network

If configs must not leave your network, run Ollama on a server rConfig can reach and use it as your provider. The Ollama tab in Set up an AI provider walks through it.

Setup at a glance

flowchart LR
    A["1 · Enable<br/>AI_CENTER_ENABLED"] --> B["2 · Accept<br/>terms"]
    B --> C["3 · Add<br/>provider"]
    C --> D["4 · Test and<br/>activate"]
    D --> E["5 · Turn on<br/>capabilities"]
    E --> F["6 · Ask your<br/>first question"]

1. Enable the AI Center on the server

   The AI Center is off by default. Add this line to /var/www/html/rconfig8/.env:

   AI_CENTER_ENABLED=true

   Then clear rConfig’s caches so it picks up the change:

   cd /var/www/html/rconfig8
   php artisan rconfig:clear-all

   Reload rConfig in your browser. AI Center now appears in the left sidebar.

   AI Center in the left sidebar after enabling the flag

2. Accept the AI Center terms

   Click AI Center. The first time you open it, rConfig shows AI Center — Terms & Conditions instead of the Overview.

   1. Click Open Terms Portal and read the terms on the rConfig portal.
   2. Back in rConfig, tick I have read and agree to the AI Center terms and conditions. (the checkbox unlocks after you open the portal link).
   3. Click I Accept.

   Acceptance is per user. Every person who uses AI features accepts once, under their own login. Until they do, AI buttons elsewhere in rConfig stay hidden for them. See Terms acceptance for details.

   AI Center Terms and Conditions page

3. Add a provider

   Go to AI Center → AI Settings and click Add Provider. Pick the provider type, give it a display name, and fill in the API key or base URL. Leave Model at its default for now. Click Add Provider to save.

   Provider-by-provider field details are in Set up an AI provider.

4. Test and activate the provider

   Saving a provider doesn’t switch it on. On the provider card:

   1. Click Test. A successful test shows Connection successful. with the model’s reply and the round-trip time.
   2. Click Activate. The card shows an Active badge.

   A tested, active provider

   The most common setup mistake

   A saved but inactive provider does nothing. If AI Settings shows the banner No AI provider is activated, click Activate on the provider you want to use.

5. Turn on capabilities

   Go to AI Center → AI Configuration and switch on the capabilities your team should use. Start with Config Analysis & Explain; it’s the quickest way to see the AI Center working. Changes apply immediately, with no save button. See AI Configuration.

6. Ask your first question

   1. Open any device with a downloaded config and view its latest configuration.
   2. Click Ask AI in the config viewer toolbar.
   3. Type “Summarise this configuration in five bullet points” and send it.

   You should get a plain-English summary within a few seconds. If you don’t, go to Troubleshooting.

   Load any device configuration to see the AI Assistant in action

Give other users access

Once it works for you, let your team in:

• Who can see the AI Center is controlled by the aicenter.view permission. Every role template includes it; check your custom roles in Roles.
• Who can change providers and capabilities needs aicenter.update, which the Admin role has. It also lets you manage other users’ MCP tokens.
• Who can create their own MCP tokens needs aicenter.mcptoken. Give it to engineers who connect agents; it doesn’t let them change AI settings.
• Each user accepts the terms the first time they open the AI Center.

Planning a team rollout?

Rolling out Iris sets out a phased plan, from pilot group to team-wide use.

Where to go next

• Use it day to day: Config Analysis & Explain, Drift Detection & Narration, Policy & Compliance Narration, and the Optimization Assistant.
• Connect Claude Code, Cursor, or another agent to your inventory with the MCP server.
• Check what data leaves your network before you roll it out: Governance, audit & privacy.