Bring Your Own AI: Custom AI Models and API Keys in CodeRoom

Flexible use of KI - tailored to your school

CodeRoom supports Bring Your Own AI by allowing the integration of custom AI sources into the platform. Schools, educational institutions, and organizations can thus determine which AI models should be used for specific functions.

Instead of relying solely on centrally provided standard models, organizations can upload their own API keys, providers, or compatible KI endpoints. CodeRoom then decides at each location whether to use the standard model or a self-configured model.

The goal is simple: more control, more flexibility, and better adaptation to the technical, organizational, and data protection requirements of each institution.

KI Modelle und Routing in CodeRoom Verwaltung

The basic principle

In CodeRoom, there are various locations where AI can be used, for example:

With Bring Your Own AI, these deployment points can be configured individually. For example, a school can specify:

The AI help for students uses its own model.
The automatic description of courses uses another model.
For certain functions, the CodeRoom standard model remains active.

This does not create a rigid either-or solution, but rather a flexible routing layer between CodeRoom and the AI providers.

Technical Approach

Technically, CodeRoom works with fixed deployment points for AI functions. These deployment points are not freely entered as text in the database, but are defined areas in the code. This ensures that new AI functions are integrated consciously and traceably.

For each deployment point, CodeRoom can resolve which model should be used:

  1. Is there a suitable user-specific configuration?
  2. Is there a suitable configuration for the institution?
  3. Should the CodeRoom standard model be used intentionally?
  4. If no specific configuration exists, the standard model is used automatically.

This architecture has an important advantage: The existing AI infrastructure remains intact. Bring Your Own AI complements the platform without invasively replacing the existing standard logic.

Supported Providers and Compatible Endpoints

Bring Your Own AI is designed to support various providers and technical variants, including:

This allows institutions to connect not only classical cloud providers, but also their own infrastructure or regional providers in the future, provided they offer a compatible interface.

API keys remain protected

The stored API keys are not stored in plain text. They are securely stored on the server-side and used only for executing the respective AI query.

This is important for schools and organizations because sensitive access data does not appear in the browser, course content, or prompt. Management remains controlled on the server-side.

In installations with multiple servers, encryption must be technically coordinated so that all servers can read the same protected keys. This is particularly relevant for larger or scaled environments.

Benefits for schools and educational institutions

More control

Institutions can decide themselves which AI models are used in which areas. This is particularly relevant if there are internal guidelines, data protection requirements, or existing contracts with specific providers.

More flexibility

Not every AI function requires the same model. A quick code help may use a different model than an extensive evaluation or administrative text generation. Bring Your Own AI enables this differentiation.

Better Cost Control

If an institution uses its own API quotas or contracts, AI costs can be controlled more directly. CodeRoom can still provide standard models, but custom models can be used when it is more sensible organizationally or economically.

Future-Proof Architecture

The AI market is developing quickly. New models, providers, and hosting variants are emerging continuously. Through the routing layer, CodeRoom remains open to new technical possibilities without each function being tied to a single provider.

Suitable for Different Scenarios

A small school can still use the standard integration. A larger educational authority can integrate custom models. An organization with special data protection requirements can use compatible custom endpoints. CodeRoom remains the same learning environment in all cases.

Example: Different AI for Different Tasks

An institution could, for example, choose the following configuration:

A fast and cost-effective model is used for direct student questions.
A stronger model is used for code evaluation.
The CodeRoom standard model remains active for administrative course descriptions.
Sensitive experiments use a custom internal endpoint.

The teacher and learning interface remains the same. The technical decision of which model is used in the background is made in the configuration.

No Obligation to Own AI

Bring Your Own AI is an extension, not a requirement. Schools can continue to use CodeRoom with the existing standard models. Those who do not want to operate or manage their own KI infrastructure need to make no changes.

This new option is mainly aimed at institutions that want more control over providers, costs, model selection, or technical infrastructure.

Conclusion

Bring Your Own AI makes CodeRoom more open and flexible. Schools and educational institutions can steer KI functions more precisely, integrate their own providers, and still use the proven CodeRoom environment.

This will make KI support in programming lessons not only more powerful but also more adaptable to the organizational realities of schools, universities, and educational institutions.