Document 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 document collection with layout-aware extraction
collection = client.collections.create(
    namespace_id="ns_your_namespace",
    name="policy_documents",
    extractors=["document-graph-extractor", "text-extractor"]
)

# Upload PDFs
client.buckets.upload(bucket_id="bkt_docs", url="s3://your-bucket/policies/")

# Retrieve relevant document sections
results = client.retrievers.execute(
    retriever_id="ret_doc_rag",
    query={"text": "What is the refund policy for enterprise customers?"}
)

# Build context with page citations
context = "\n".join([
    f"[{i+1}] {doc['text']} (Document: {doc['root_object_id']}, Page {doc['page_number']})"
    for i, doc in enumerate(results["documents"])
])

# Generate answer
response = openai.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": f"Answer using these document excerpts:\n{context}"},
        {"role": "user", "content": "What is the refund policy for enterprise customers?"}
    ]
)
print(response.choices[0].message.content)