nurricai 0.1.0

Python client for the ACRA API — Adaptive Contextual Retrieval Architecture

nurricai

Python client for the ACRA API — Adaptive Contextual Retrieval Architecture by NurricAI.

ACRA is intelligent RAG-as-a-service. It automatically classifies your query complexity and adapts retrieval depth — simple questions get fast single-hop retrieval, complex multi-hop questions get deeper iterative retrieval with Gemma 27B generation.

Install

pip install nurricai

Quickstart

from nurricai import ACRA

acra = ACRA(api_key="nacra-your-key-here")

# Index your documents
acra.ingest(["ACRA stands for Adaptive Contextual Retrieval Architecture..."])

# Query
response = acra.query("What does ACRA stand for?")
print(response.answer)
print(f"Credits used: {response.credits_used}")
print(f"Credits left: {response.credits_remaining}")

With metadata and namespaces

acra.ingest(
    texts=["Page 1 content...", "Page 2 content..."],
    metadata=[{"source": "manual.pdf", "page": 1},
              {"source": "manual.pdf", "page": 2}],
    namespace="my-project"
)

response = acra.query("What is on page 2?", namespace="my-project")
for source in response.sources:
    print(source.metadata, source.score)

Error handling

from nurricai import ACRA, OutOfCreditsError, AuthError, ACRAError

try:
    response = acra.query("What is X?")
except OutOfCreditsError as e:
    print(f"Out of credits. Upgrade at {e.upgrade_url}")
except AuthError:
    print("Invalid API key")
except ACRAError as e:
    print(f"Error: {e}")

Plans

Plan	Credits	Price
Free	100 / day	$0
Hobby	5,000 / month	$10
Team	25,000 / month	$50
Business	100,000 / month	$100
Enterprise	Unlimited	$499

Get your API key at nurricai.com