jschm42 / talkforgeai

TalkforgeAI - AI-Powered Chatbot Creation Tool

TalkforgeAI - AI-Powered Chatbot Creation Tool

Overview

TalkforgeAI is an advanced AI-powered chat software designed to facilitate the creation of highly personalized chatbots, now referred to as "Assistants". This software integrates with multiple AI platforms including OpenAI, Mistral, Anthropic, and Ollama, supporting both cloud-based and local AI models. Users can customize each Assistant with unique personality traits and an avatar image. The chatbots can interact via text or voice inputs, with responses delivered through browser-based or Elevenlabs text-to-speech output.

Note: This project is still in the early stages of development and may not yet be suitable for production use.

Prerequisites

• An API key from any supported AI provider (OpenAI, Mistral, Anthropic, Ollama) is required.
• For optional Elevenlabs speech output, an API key is also necessary.

Features

• 🤖 Create and manage multiple, GPT-style, chatbot Assistants with custom personalities using AI providers like OpenAI, Mistral, Anthropic, and Ollama.
• 🎨 Each assistant can be customized with unique personality traits, an avatar image and one of the supported AI models.
• ☁️ Support for both cloud-based and local AI models, enhancing flexibility and control over data processing.
• 🗣️ Support for text and voice input, with text-to-speech output through Elevenlabs or the browser build-in MS Speech API.
• 🖼️ Generate a profile picture for your Assistant directly inside the app.
• 🎭 Image generation through DALL-E 3, compatible with various GPT models.
• 💻 Code highlighting for generated source code.
• 📦 Docker container support for simplified deployment.
• 🧠 Memory Feature with Vector Databases: TalkforgeAI supports persistent memory capabilities for storing essential information. This allows your Assistants to remember past interactions and provide more personalized responses. Currently only local storage is supported, other databases are in development.

Upcoming Features

• Enhanced multi-user support with comprehensive access controls for managing Assistants.
• Support for uploading and interpreting files (PDF, TXT, etc.).
• Persistent memory capabilities for storing essential information.
• Support for Vector Storage like Apache Cassandra, Redis, Neo4j and local storage.
• Expanded task functions like email automation and calendar queries.
• Integration with additional AI providers and services.
• Enhanced image generation capabilities with OpenAI and StablilityAI .
• Advanced diagram creation with and graphical enhancement of LaTeX code.

Installation

Building the Project

• Ensure Java 21 or higher is installed on your system and the JAVA_HOME environment variable is correctly set.

Starting the Server

To run from a JAR file, include API keys and specify the data directory as shown in this example:

java -jar ./talkforgeai-0.1.0.jar --server.port=8090 --spring.ai.openai.api-key=[your API Key] --talkforgeai.datadir=[your TalkforgeAI data directory]

Alternatively, configure using a properties file:

java -jar ./talkforgeai-0.1.0.jar --spring.config.additional-location=./talkforgeai.properties

Example properties file (if you don't want to use a model, leave property empty or remove it from the file):

spring.ai.openai.api-key=[your API Key]
spring.ai.mistralai.api-key=[your API Key]
spring.ai.anthropic.api-key=[your API Key]
spring.ai.ollama.base-url=http://localhost:11434
elevenlabs.api-key=[your API Key]
server.port=8090
talkforgeai.datadir=[your TalkforgeAI data directory]

Available properties:

Property	Default Value	Description
server.port	8090	The port on which the server will run.
talkforgeai.datadir	${user.home}/.talkforgeai	The directory where TalkforgeAI data will be stored.
spring.servlet.multipart.max-file-size	5MB	The maximum file size for multipart file uploads.
spring.servlet.multipart.max-request-size	5MB	The maximum request size for multipart file uploads.
spring.ai.openai.api-key	-	The API key for OpenAI.
spring.ai.mistralai.api-key	-	The API key for Mistral AI.
spring.ai.ollama.base-url	http://localhost:11434	The base URL for Ollama AI.
spring.ai.anthropic.api-key	-	The API key for Anthropic AI.
spring.ai.anthropic.version	2023-06-01	The version of Anthropic AI.
spring.ai.stabilityai.api-key	-	The API key for StabilityAI.
elevenlabs.api-key	-	The API key for Elevenlabs.
elevenlabs.api-url	https://api.elevenlabs.io	The base URL for Elevenlabs API.
spring.datasource.url	jdbc:sqlite:${talkforgeai.datadir}/talkforgeai.db	The URL for the SQLite database.
spring.datasource.driverClassName	org.sqlite.JDBC	The driver class name for the SQLite database.
spring.jpa.database-platform	org.hibernate.community.dialect.SQLiteDialect	The dialect for the SQLite database.
logging.level.com.talkforgeai	INFO	The log level for TalkforgeAI.

Using Docker

You can build and run the Docker container locally using the Dockerfile provided in the project. Here are the steps:

1. Build the Docker image. Navigate to the project root directory where the Dockerfile is located and run the following command:

   docker build -t talkforgeai:latest .

   This command builds a Docker image using the Dockerfile in the current directory and tags the image as talkforgeai:latest.

2. Create an .env file on your system where you'll specify all necessary environment variables for the Docker container. Example content for your .env file:

   SPRING_AI_OPENAI_API_KEY=[your OpenAI API Key]
   SPRING_AI_MISTRALAI_API_KEY=[your Mistral AI API Key]
   SPRING_AI_ANTHROPIC_API_KEY=[your Anthropic API Key]
   SPRING_AI_STABILITYAI_API_KEY=[your StabilityAI API Key]
   SPRING_AI_OLLAMA_BASE_URL=http://localhost:11434
   ELEVENLABS_API_KEY=[your ElevenLabs API Key]
   

3. Run the Docker container using the .env file to provide the environment variables:

   docker run -d -p [your local port]:8090 --env-file /path/to/your/.env file talkforgeai:latest

   Replace /path/to/your/.env with the actual path to your .env file and [your local port] with the port number you want to use on your host.

   For example, if your .env file is located at /home/user/talkforgeai/.env and you want to run your container on local port 8090, the command would be:

   docker run -d -p 8090:8090 --env-file /home/user/talkforgeai/.env talkforgeai:latest

This approach allows you to manage environment variables centrally within the .env file, simplifying the Docker command and enhancing security by avoiding the direct listing of sensitive information in the command line or scripts.

Building from Source

After cloning the repository, build the project using Maven:

mvn clean package

Note: Replace placeholder text (e.g., [your API Key]) with actual API keys and directory paths as needed.

Contributing

Contributions are welcome! Review our contributing guidelines to get started.

License

TalkforgeAI is released under the Apache License 2.0. See the LICENSE file for more details.

Dependencies and Licenses

This project uses several open-source libraries as dependencies. The direct dependencies and their licenses are listed in the DEPENDENCIES file.

Please note that these dependencies may have their own transitive dependencies with various licenses. Users of this project should be aware of these licenses when using the software.

To view a full list of dependencies and their licenses, you can use the following Maven command:

mvn license:aggregate-add-third-party

Support

For support and inquiries, please open an issue on the GitHub repository or contact the maintainers directly.

talkforgeai's Issues

Missing transformations for markup code

There are some missing transformations for markus code like Bold text or ### Header.

There is a problem with some dto files.

Assistant profile picture bug

Profile picture is shown corretly:

After switching to another GPT:

This is wrong. The profile picture of a message must always be the one its generated with.