AI API

The AI API provides endpoints for managing AI models, inference, training, and advanced AI operations.

Status: Roadmap

This API is on the development roadmap. The endpoints documented below represent the planned interface design.

Base URL

http://localhost:9000/api/v1/ai

Authentication

Uses the standard botserver authentication mechanism with appropriate role-based permissions.

Endpoints

Model Management

Inference

Training

Model Configuration

Request Examples

List Available Models

models = GET "/api/v1/ai/models"
FOR EACH model IN models
    TALK model.name + " - " + model.status
NEXT

Chat Completion

request = NEW OBJECT
request.model = "gpt-4"
request.messages = NEW ARRAY
request.messages.ADD({"role": "user", "content": "Hello, how are you?"})

response = POST "/api/v1/ai/chat/completions", request
TALK response.choices[0].message.content

Generate Embeddings

request = NEW OBJECT
request.input = "Convert this text to embeddings"
request.model = "text-embedding-3-small"

result = POST "/api/v1/ai/embeddings", request
embedding = result.data[0].embedding

Start Training Job

training_config = NEW OBJECT
training_config.base_model = "llama-2-7b"
training_config.dataset = "my-training-data"
training_config.epochs = 3

job = POST "/api/v1/ai/training/start", training_config
TALK "Training job started: " + job.id

Response Codes

Required Permissions