Meez
AI Assistant and Copilot SDK for SaaS Applications
Meez is a powerful Python library that provides AI-powered assistance capabilities for SaaS applications. It offers intention detection, contextual responses, and workflow automation using LangChain and LangGraph.
Features
Intention Detection: Automatically detect user intentions from natural language input
Contextual Responses: Generate intelligent responses based on provided context data
Workflow Automation: Create complex AI workflows using LangGraph
Multiple Data Sources: Support for text, JSON, and file-based data sources
OpenAI Integration: Built on top of OpenAI’s GPT models for reliable AI responses
Flexible Architecture: Modular design for easy integration and customization
Installation
Install Meez using pip:
pip install Meez
Quick Start
Set up your OpenAI API key:
import os
os.environ["OPENAI_API_KEY"] = "your-openai-api-key-here"
Basic intention detection:
from meez.core import Langchain, Intention
# Initialize LangChain
langchain = Langchain(openai_api_key="your-api-key")
# Define possible intentions
intentions = ["greeting", "weather", "help", "goodbye", "unknown"]
# Create intention detector
detector = Intention(langchain=langchain, intentions=intentions)
# Detect intention
user_input = "Hello there!"
intention = detector.detect(user_input)
print(f"Detected intention: {intention}")
Contextual responses:
from meez.core import Langchain, Respond
from meez.data import TextReader
# Initialize components
langchain = Langchain(openai_api_key="your-api-key")
respond = Respond(langchain)
# Create data source
context_data = "Python is a high-level programming language..."
data_reader = TextReader(context_data)
# Get contextual response
question = "What is Python?"
response = respond.run(question=question, data=data_reader)
print(f"Response: {response}")
Usage Examples
Intention Detection
Detect user intentions from natural language:
import os
from meez.core import Langchain, Intention
# Setup
api_key = os.getenv("OPENAI_API_KEY")
langchain = Langchain(openai_api_key=api_key)
# Define intentions
intentions = [
"greeting",
"weather",
"joke",
"help",
"goodbye",
"book_appointment",
"cancel_appointment",
"unknown"
]
# Create detector
detector = Intention(langchain=langchain, intentions=intentions)
# Test inputs
test_texts = [
"Hello there!",
"What's the weather like?",
"I need to book an appointment",
"Can you help me cancel my appointment?"
]
for text in test_texts:
intention = detector.detect(text)
print(f"'{text}' → {intention}")
Contextual Responses
Generate responses based on context data:
from meez.core import Langchain, Respond
from meez.data import TextReader, JsonReader, FileReader
# Initialize
langchain = Langchain(openai_api_key=api_key)
respond = Respond(langchain)
# Using text data
text_data = "Artificial Intelligence is a branch of computer science..."
text_reader = TextReader(text_data)
response = respond.run(
question="What is AI?",
data=text_reader
)
# Using JSON data
json_data = {"company": {"name": "TechCorp", "employees": 250}}
json_reader = JsonReader(json_data)
response = respond.run(
question="How many employees does the company have?",
data=json_reader
)
# Using file data
file_reader = FileReader("documentation.txt")
response = respond.run(
question="What are the main features?",
data=file_reader
)
General Respond
Respond to general user questions using AI knowledge:
from meez.core import Langchain, GeneralRespond
# Initialize
langchain = Langchain(openai_api_key=api_key)
respond = GeneralRespond(langchain)
# Get response
question = "What is the capital of France?"
response = respond.run(question=question)
print(f"Response: {response}")
Workflow Automation with LangGraph
Create complex AI workflows:
from meez.core.langgraph import LangGraph, MainState
from meez.core import Langchain, Intention
# Define workflow nodes
def get_intent(state: MainState) -> MainState:
# Detect user intention
detector = Intention(langchain, ["get_phone", "get_email", "unknown"])
intent = detector.detect(state["messages"][-1]["content"])
state["messages"].append({"role": "assistant", "content": intent, "internal": True})
return state
def decide(state: MainState) -> str:
# Return the detected intent to determine next step
return state["messages"][-1]["content"]
def get_phone(state: MainState) -> MainState:
state["messages"].append({"role": "assistant", "content": "Phone: +1234567890"})
return state
def get_email(state: MainState) -> MainState:
state["messages"].append({"role": "assistant", "content": "Email: support@company.com"})
return state
def unknown(state: MainState) -> MainState:
state["messages"].append({"role": "assistant", "content": "I'm sorry, I don't know that."})
return state
# Create and configure graph
graph = LangGraph()
graph.add_node("get_intent", get_intent)
graph.add_node("decide", decide)
graph.add_node("get_phone", get_phone)
graph.add_node("get_email", get_email)
graph.add_node("unknown", unknown)
graph.set_entry_point("get_intent")
graph.add_conditional_edge("get_intent", decide)
graph.add_finish_point("get_phone")
graph.add_finish_point("get_email")
graph.add_finish_point("unknown")
# Run workflow
initial_state = {"messages": [{"role": "user", "content": "What's your phone number?"}]}
result = graph.run(initial_state)
Examples
See the examples/ directory for complete working examples:
intention_detection.py - Basic intention detection
respond_to_user.py - Contextual responses with different data sources
complex_graph.py - Advanced workflow automation
sample_graph.py - Simple graph workflow
sample_assistant.py - Complete assistant implementation
Development
Setup Development Environment
# Clone the repository
git clone https://github.com/clivern/meez.git
cd meez
# Install development dependencies
pip install -r requirements.test.txt
pip install -e .
Running Tests
# Run tests
make ci
Support
Documentation: https://github.com/clivern/meez/
Email: hello@clivern.com
Changelog
See CHANGELOG.rst for a detailed history of changes.
Metadata
Release files for Meez 0.3.4
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| meez-0.3.4.tar.gz | 25.9 kB | Without provenance | Details for meez-0.3.4.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| meez-0.3.4-py3-none-any.whl | Python 3 | none | any | Details for meez-0.3.4-py3-none-any.whl |
Total release size: 45.9 kB
Release files / meez-0.3.4.tar.gz
| Download URL | meez-0.3.4.tar.gz |
|---|---|
| Size | 25.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
0e2d4e56235995ebe1fb22c945cb08811a6181ffb0ceac6f2c1d757aa12a5713
|
|
BLAKE2b-256 checksum How to use checksums |
d0c207b99a00524494fb4ef7c95f1d02aa1943cb4a21ede7b87c162289422986
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.9.23
|
Release files / meez-0.3.4-py3-none-any.whl
| Download URL | meez-0.3.4-py3-none-any.whl |
|---|---|
| Size | 20.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1103d68b07c5c54ef9870a8450951d8d13914aac8929e2ba3eae0665dd73fc96
|
|
BLAKE2b-256 checksum How to use checksums |
2aec38d66a78ab80fca405395018d06ac54851ad3b4093b8dcc05a4864089f91
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.9.23
|
Security for Meez 0.3.4
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.