conduit-ai 0.2.0

Python SDK for Conduit — the knowledge graph engine. Includes LangChain retriever.

conduit-ai

Python SDK for Conduit — the knowledge graph engine.

Install

pip install conduit-ai                    # Core client
pip install 'conduit-ai[langchain]'       # With LangChain retriever

Quick Start

Ask a question

from conduit_ai import ConduitClient

client = ConduitClient(api_key="ck_...", endpoint="http://localhost:4000")

answer = client.ask("How does Snowflake Cortex Search work?")
print(answer.answer)
print(f"Sources: {len(answer.sources)}")

Conversational follow-ups

import uuid

thread_id = str(uuid.uuid4())

answer1 = client.ask("What is Delta Live Tables?", thread_id=thread_id)
answer2 = client.ask("Can I use it with Cortex?", thread_id=thread_id)
# ^ automatically rewritten to: "Can I use Delta Live Tables with Snowflake Cortex?"

Retrieve context (no LLM)

ctx = client.context("data pipeline best practices", limit=5)
for result in ctx.results:
    print(f"{result.title} ({result.score:.2f})")
    print(f"  Domains: {result.domains}")

Stream tokens

async for token in client.aask_stream("Compare Databricks and Snowflake for ML"):
    print(token, end="", flush=True)

LangChain Retriever

Drop Conduit into any LangChain/LangGraph chain:

from conduit_ai.retriever import ConduitRetriever
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser

retriever = ConduitRetriever(
    api_key="ck_...",
    endpoint="http://localhost:4000",
    kai_id="kai_snowflake",   # Optional: scope to a Kai
    limit=8,
)

prompt = ChatPromptTemplate.from_template(
    "Answer based on context:\n{context}\n\nQuestion: {question}"
)

chain = (
    {"context": retriever, "question": RunnablePassthrough()}
    | prompt
    | ChatOpenAI(model="gpt-4o")
    | StrOutputParser()
)

result = chain.invoke("How do I set up change data capture?")

Retriever Options

Parameter	Default	Description
api_key	None	Conduit API key
endpoint	https://api.conduit.datakai.com	Conduit server URL
kai_id	None	Scope retrieval to a specific Kai
limit	8	Max results to retrieve
include_content	True	Include full zettel content in documents
include_graph_context	False	Use /context endpoint (includes graph neighbors) instead of /ask

Scoped Knowledge (Kais)

Kais are knowledge views — filtered subsets of the graph scoped by domain, topic, or knowledge type.

# Query only Snowflake knowledge
client = ConduitClient(api_key="ck_...", kai_id="kai_snowflake")
answer = client.ask("How do dynamic tables work?")

# Or per-retriever
retriever = ConduitRetriever(api_key="ck_...", kai_id="kai_aws")

Async Support

Every method has an async counterpart:

answer = await client.aask("question")
ctx = await client.acontext("query")

async with ConduitClient(api_key="ck_...") as client:
    answer = await client.aask("question")