ai-systems/system-router

System Router

System Router is a lightweight reference text-classification model for routing AI-system design requests into practical architecture categories.

It is published under the ai-systems handle as a small, transparent demonstration model.

Routing Labels

• agents
• control
• interoperability
• knowledge
• memory
• orchestration
• physical-ai
• validation

Example

Input:

Use a human approval step before an agent can deploy code.

Expected route:

control

Input:

Route hard reasoning tasks to a stronger model.

Expected route:

orchestration

Usage

from joblib import load

router = load("system-router.joblib")

text = "Store long-term memories for a persistent agent"
prediction = router.predict([text])[0]

print(prediction)

Model Architecture

The reference model uses:

• TF-IDF text features
• unigram and bigram features
• logistic regression classification

The model is intentionally small so the full routing approach remains easy to inspect and reproduce.

Training Data

The model was trained on a small curated set of AI-system architecture prompts covering:

• model routing
• agent execution
• memory
• permissions and control
• validation
• interoperability
• Physical AI
• retrieval and knowledge systems

Intended Use

Suitable for:

• demos
• architecture explorers
• lightweight routing experiments
• educational examples
• prototyping

Limitations

This is a reference model, not a production-grade router.

It was trained on a small curated dataset, so predictions outside the covered architecture domains may be unreliable. It should not be used for safety-critical, financial, legal, medical, or other high-impact decisions.

Related Dataset

ai-systems/ai-system-patterns

License

Apache-2.0