Video RAG Pipeline

from mixpeek import Mixpeek
from openai import OpenAI

client = Mixpeek(api_key="YOUR_API_KEY")
openai = OpenAI(api_key="YOUR_OPENAI_KEY")

# Create video collection with scene + transcript extraction
collection = client.collections.create(
    namespace_id="ns_your_namespace",
    name="training_videos",
    extractors=["multimodal-extractor", "text-extractor"]
)

# Retrieve relevant video segments
results = client.retrievers.execute(
    retriever_id="ret_video_rag",
    query={"text": "How do I configure the firewall settings?"}
)

# Build context with video citations
context = "\n".join([
    f"[{i+1}] {doc['text']} (Video: {doc['root_object_id']}, {doc['start_time']:.1f}s-{doc['end_time']:.1f}s)"
    for i, doc in enumerate(results["results"])
])

# Generate answer with citations
response = openai.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": f"Answer using this video context:\n{context}"},
        {"role": "user", "content": "How do I configure the firewall settings?"}
    ]
)
print(response.choices[0].message.content)