Skip to content

Latest commit

 

History

History

multi-turn-agents

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 
app.py
app.py
 
 

README.md

🧠 Multi-Turn-Agents in an Application

This repository contains a Streamlit web application that leverages Tune Studio, an advanced platform for building and managing AI assistants using multiple large language models (LLMs), to conduct market research and generate marketing campaigns. The app uses different AI assistants to:

  1. Extract and analyze market research data.
  2. Generate a detailed image prompt using the Claude Sonet model.
  3. Produce the final campaign poster image using GPT-4o for image generation.

The application enables you to:

  • Perform market research.
  • Analyze insights from the data.
  • Generate a marketing campaign poster.
  • Download the results as a PDF.

🎯 Features

  • Market Research Assistant: Query the AI assistant to conduct in-depth market research using the models available on Tune Studio.
  • Campaign Prompt Generator: Analyze research data to generate an image prompt that aligns with the research insights.
  • Campaign Poster Generator: Create an image prompt to generate a marketing poster using GPT-4o.
  • Download PDF: Export all the research, analysis, and generated content as a PDF.

🛠️ Installation

To run this application locally, follow the steps below:

1. Clone the Repository

git clone https://github.com/aryankargwal/genai-tutorials.git
cd genai-tutorials/multi-turn-agents

2. Set up API Key

The application requires an API key to interact with Tune Studio for the AI assistants. During runtime, you can input this key into the app's sidebar.

3. Run the Application

streamlit run app.py

The app should now be running on http://localhost:8501/.

🔑 API Key

You’ll need an API key from Tune Studio to use their AI assistant models. Tune Studio offers access to a variety of models such as Claude Sonet, GPT-4o, LLaMA 3.1, Haiku, and more. You can input your API key in the app's sidebar during usage.

🧩 Application Workflow

1. Market Research:

  • Enter a query to conduct market research using Tune Studio’s AI assistant (based on a model like Claude Sonet or LLaMA 3.1).
  • The AI assistant retrieves relevant research and insights from the web.
  • Results are displayed under the "Market Research" section.

2. Generate Campaign Prompt:

  • Once research is performed, the assistant analyzes the insights and generates a prompt for image creation.
  • This prompt is crafted to fit the style and tone of the marketing campaign you are aiming for.
  • Results are displayed under "Suggested Prompt."

3. Generate Campaign Poster:

  • Using the image generation prompt, the app interacts with the GPT-4o model to generate a campaign poster image.
  • The generated image description is displayed under "Generated Campaign."

4. Download PDF:

  • You can export the market research, analysis, and campaign outputs as a PDF file.

🧑‍💻 Code Overview

Main Functions

  • call_market_research_assistant(query): Interacts with the market research AI assistant via Tune Studio to fetch data based on the user’s query.
  • call_analytics_assistant(research_text): Analyzes research data to generate an image generation prompt for the campaign poster.
  • call_image_generation(analysis_text): Uses the AI assistant (GPT-4o) to generate the marketing campaign poster.
  • create_pdf(market_research, analysis_result, image_output): Compiles the market research, analysis, and image output into a PDF.

Key Libraries

  • Streamlit: For the web application interface.
  • Requests: To handle API calls to the Tune Studio AI assistants.
  • FPDF: For generating downloadable PDF reports.

🌐 Tune Studio Platform

Tune Studio allows you to create and manage AI assistants on their platform, offering multi-functionality for large language models such as Claude Sonet, GPT-4o, LLaMA 3.1, Haiku, and more. This application utilizes these models to perform tasks like:

  • Market research (e.g., using Claude Sonet or LLaMA 3.1).
  • Image generation (e.g., GPT-4o).
  • Text-based analysis for generating image prompts.

Tune Studio simplifies access to multiple models through API calls, allowing flexible integration into applications like this one.

📝 License

This project is licensed under the MIT License.

🤝 Contributing

Feel free to submit issues or pull requests if you have suggestions or improvements for the application.

🌟 Acknowledgments

  • Streamlit for providing an easy-to-use platform for building interactive web apps.
  • Tune Studio for enabling seamless integration with various large language models and multi-functional AI assistants.