Savantly base-bot

This is a chat bot based on the LangChain chat bot, packaged up and parameterized for easy re-use.
The app leverages LangChain's streaming support and async API to update the page in real time for multiple users.

You can run the docker container with mounted volumes to modify the default behavior.

TLDR;

Run in docker (requires your OPENAI_API_KEY as an env variable)

docker run -p 9000:9000 -e OPENAI_API_KEY=${OPENAI_API_KEY} savantly/base-bot

Or deploy the Helm Chart chart info

Helm must be installed to use the charts. Please refer to Helm's documentation to get started.

Once Helm has been set up correctly, add the repo as follows:

helm repo add base-bot https://savantly-net.github.io/base-bot/

If you had already added this repo earlier, run helm repo update to retrieve the latest versions of the packages. You can then run helm search repo base-bot to see the charts.

To install the base-bot chart:

helm install my-base-bot base-bot/base-bot

To uninstall the chart:

helm delete my-base-bot

Customizing

The easiest way to get started is to copy the docker-compose.yml to your project folder and create a data directory for your own domain.
You can get starter data here - ./data
The do docker compose up .

Override other settings as needed.

✅ Developing

Install dependencies: pip install -r requirements.txt
Run ingest.sh to ingest LangChain docs data into the vectorstore (only needs to be done once).
1. You can use other Document Loaders to load your own data into the vectorstore.
Run the app: make start
Open localhost:9000 in your browser.

Configuration

Modify the settings.py to customize the chat parameters.
Add a .secrets.py file to hold your secret data.

Both of these files can be mounted in place if you're using the pre-built Docker image.

Environment Variables

If you're running the docker image, you can easily customize the configurations by using environment variables.

Prefix the configuration items with SAVANTLY_CHAT_ to pass env vars directly to the base-bot configuration.

For example -

SAVANTLY_CHAT_UI_PAGE_TITLE="My Bot"

LLM Configs

Use these parameters to modify the LLM configuration

LLM_REPRHASING_MODEL = "gpt-3.5-turbo"
LLM_REPHRASING_TEMPERATURE = 0.0
LLM_REPHRASING_VERBOSE = True
LLM_STREAMING_MODEL = "gpt-3.5-turbo"
LLM_STREAMING_TEMPERATURE = 0.0
LLM_STREAMING_VERBOSE = True

White labeling

Use these settings to customize the branding -

UI_PAGE_TITLE = "Savantly Base Bot"
UI_PAGE_DESCRIPTION = "Savantly Base Bot"
UI_PAGE_FAVICON = "https://savantly.net/img/favicon.png"
UI_SHOW_HEADER = True
UI_HEADER_CENTER = True
UI_HEADER_TITLE = "BASEBOT"
UI_HEADER_LOGO_SRC = "https://savantly.net/img/logo.png"
UI_HEADER_LOGO_ALT = "Savantly Logo"
UI_HEADER_LOGO_HREF = "https://savantly.net"
UI_CHAT_BOT_NAME = "BaseBot"

📚 Technical description

There are two components: ingestion and question-answering.

Ingestion has the following steps:

Scan the ./data/docs folder for content to load. The scan ignores any files that begin with a dot.
Load the documents with LangChain's DirectoryLoader.
Split documents with LangChain's TextSplitter
Create a vectorstore of embeddings, using LangChain's vectorstore wrapper (with OpenAI's embeddings and FAISS vectorstore).

Question-Answering has the following steps

Given the chat history and new user input, determine what a standalone question would be (using GPT-3).
Given that standalone question, look up relevant documents from the vectorstore.
Pass the standalone question and relevant documents to GPT-3 to generate a final answer.

Chatwoot Integration

An endpoint is provided for Chatwoot webhook integration.
This allows Base-Bot to reply as an "Agent Bot" for any Chatwoot inbox.
Enable it with the setting CHATWOOT_ENABLED = True

Additional Settings

CHATWOOT_URL = "https://chatwoot.com"
CHATWOOT_BOT_TOKEN = ""
CHATWOOT_BOT_VARIANT = ""

The webhook URL for Base-Bot is /chatwoot/bot

Roadmap

Parameterize prompts and ingestion settings
Enable loading docs from a folder
Create Docker image
Create docker-compose for rapid testing
Create parameters and code to substitute the pickled vector store for Pinecone or another implementation
Create Helm chart
Create web-component for dropping the chat on any web page
Parameterize additional chain settings??

savantly-net/base-bot