README

This repository contains a Flask application that demonstrates how to use Temporal and OpenAI to build a simple chatbot. The chatbot is implemented using a workflow that uses an activity to call the OpenAI API to generate responses to user questions.

Running the application

Prerequisites:

• Python >= 3.7
• Poetry
• Local Temporal server running
• Open AI keys
  • Create an .env file and save your key as OPENAI_API_KEY="$SecretKEY.

With this repository cloned, run the following at the root of the directory:

poetry install

1. Start the Temporal server by running temporal server start-dev.
2. In a new terminal window, start the worker by running poetry run python run_worker.py
3. In another terminal window, start the Flask application by running python run_workflow.py.
4. Open your web browser and go to http://localhost:5000.

UI

Interact with the model with the Flask UI.

See the results from Temporal's Web UI at: http://localhost:8233/.

Files

your_activities.py

This file defines an activity called return_answer, which calls the OpenAI API to generate responses to user questions. It takes an instance of a OpenAIQuestion dataclass, which contains the user's question, and returns a string containing the response.

your_openai.py

This file defines a workflow called OpenAIWorkflow. The run method takes a question string as input and calls the return_answer activity to generate a response. The workflow returns the response as a string.

run_workflow.py

This file defines a Flask application that serves as a user interface for the chatbot. When the user submits a question, the application creates an instance of OpenAIWorkflow and executes it using the Temporal client. The application then displays the user's question and the response generated by the chatbot.

run_worker.py

This file starts the Temporal worker, which listens for and executes workflows and activities. It uses the return_answer activity and the OpenAIWorkflow workflow.