Back to licensing

AI Center: What Network Engineers Need to Know

Operational Safety Summary
Aligned with AI Center Terms Version 1.1
Last Updated: 29th August 2026

This summary explains the main operational responsibilities and safety requirements for network engineers using rConfig AI Center.

It does not replace the rConfig AI Center Terms of Use Addendum or the applicable rConfig License Agreement. If this summary conflicts with either agreement, the relevant agreement controls.

The Short Version

AI Center can help you understand configurations, investigate issues, draft commands and prepare changes. It is an assistive tool, not a network engineer, change approver or autonomous network management system.

You remain responsible for deciding what data is sent, checking every response and safely applying any resulting change.

Prompts, configurations, operational context and AI responses travel directly between your local rConfig installation and the AI provider or model selected by your organisation. They never pass through any rConfig service.

1. Your Organisation Chooses the AI Provider and Model

Your organisation selects, connects and controls the AI provider, hosted service or self-hosted model used by AI Center.

rConfig does not choose the provider, operate the model or control how the provider behaves. The availability and quality of responses may change when the provider updates or replaces a model.

2. Your Data Goes Directly to Your Provider

Prompts, configurations, device information, operational context and AI responses pass directly and exclusively between your local rConfig installation and the configured provider endpoint.

They are never received, routed, proxied, stored, cached, logged or inspected by any rConfig operated service.

rConfig will not change this architecture under the AI Center Terms. Any future service using a different data path would have to be separate, optional, separately agreed and explicitly enabled by an Authorised Administrator.

3. You Control What Is Sent

You decide what information is submitted to the AI provider. Before sending anything, check for:

  1. Passwords, API keys, tokens, certificates and private keys.
  2. Personal data or regulated information.
  3. Confidential business or customer information.
  4. Export controlled or third party proprietary information.

Remove, mask or minimise sensitive information before transmission. Do not assume AI Center or the provider will find and remove every secret.

4. Your Provider's Rules Apply

Your organisation is responsible for understanding and configuring the provider's rules for:

  1. Data storage and deletion.
  2. Use of submitted data for model training.
  3. Human review by the provider.
  4. Processing locations and international transfers.
  5. Usage limits, token charges and other fees.

These matters are governed by your organisation's direct agreement with its selected provider, not by rConfig.

5. AI Output Can Be Wrong

AI responses are generated by a model and are not independently verified by rConfig. A response may:

  1. Invent commands, features or configuration syntax.
  2. Misunderstand the device, vendor, topology or operational context.
  3. Omit dependencies, risks, side effects or rollback steps.
  4. Recommend insecure, destructive or unsupported actions.
  5. Produce a different answer when the same question is asked again.

Treat every response as an unverified draft, not authoritative engineering guidance.

6. Review Everything Before Use

Before using AI generated commands, scripts, configurations or recommendations:

  1. Confirm the source data and device scope.
  2. Check the syntax and intended effect.
  3. Compare the response with authoritative vendor documentation.
  4. Check it against your organisation's standards and security policies.
  5. Obtain the normal technical and change approvals.

AI Center is not a change approval authority. Your approval means you accept responsibility for the proposed action.

7. Test Before Production

Use the same safeguards you would apply to any manually prepared change:

  1. Test in a non-production or controlled environment where possible.
  2. Maintain a current backup.
  3. Prepare and test a rollback or recovery procedure.
  4. Use an approved maintenance window.
  5. Monitor execution and confirm the resulting network state.

Do not bypass normal change control because a command or configuration was generated by AI.

8. Do Not Automatically Apply Unreviewed Output

AI generated output must not be applied automatically to production infrastructure without qualified human review and approval.

After approval, you may use available rConfig features to schedule or automate execution. Automation does not remove responsibility for scope, testing, permissions, monitoring, rollback or consequences.

9. Protect Provider Credentials

API keys, tokens, certificates and other provider credentials remain within your local environment and are never transmitted to or held by rConfig.

You must:

  1. Give credentials only the permissions they need.
  2. Restrict who can add or change provider endpoints and credentials.
  3. Restrict outbound access to approved provider endpoints.
  4. Monitor provider usage, billing and account activity.
  5. Rotate or revoke credentials if compromise is suspected.

10. Watch for Prompt Injection and Malicious Context

Configuration data, retrieved documents or user supplied text may contain hidden or misleading instructions designed to manipulate the model.

AI Center and the selected model cannot detect every prompt injection, poisoned input, unsafe command or data exfiltration attempt.

Use least privilege access, control trusted data sources, review unusual output and disable the affected provider connection or credentials if unsafe behaviour is suspected.

Before You Use AI Center

Confirm that:

  1. You know which provider and model are connected.
  2. You understand what data may leave your local environment for that provider.
  3. You have checked the provider's storage, training and privacy settings.
  4. You will remove or mask sensitive information before submission.
  5. You will independently review and test any output before use.
  6. You will follow normal approval, backup and rollback procedures.

Operational Acknowledgement

By acknowledging this summary, you confirm that you have read and understood the operational safety requirements for using AI Center.

This acknowledgement is not contractual acceptance of the AI Center Terms of Use Addendum and does not confirm authority to bind your organisation. Contractual acceptance must be completed separately by an Authorised Administrator through the required rConfig acceptance workflow.

We use cookies to improve this site. Cookie policy