TdAi

To start your Phoenix server:

• Run mix setup to install and setup dependencies
• Start Phoenix endpoint with mix phx.server or inside IEx with iex -S mix phx.server

Now you can visit localhost:4000 from your browser.

Ready to run in production? Please check our deployment guides.

Learn more

• Official website: https://www.phoenixframework.org/
• Guides: https://hexdocs.pm/phoenix/overview.html
• Docs: https://hexdocs.pm/phoenix
• Forum: https://elixirforum.com/c/phoenix-forum
• Source: https://github.com/phoenixframework/phoenix

docker build --build-arg APP_NAME=td_ai --build-arg APP_VERSION=0.0.1 --volume /home/bluetab/Escritorio/Truedat/td-ai:/app .

docker build --build-arg APP_VERSION=0.1.0 .

Environment variable

• VAULT_TOKEN Token for Vault connection
• VAULT_SECRETS_PATH Prefix path for vault secrets
• REDIS_AUDIT_STREAM_MAXLEN (Optional) Maximum length for Redis audit stream. Default: 100
• REDIS_STREAM_MAXLEN (Optional) Maximum length for Redis stream. Default: 100

Oban configuration

• OBAN_DB_SCHEMA (Optional) Schema prefix to use for Oban tables. Default: private.
• OBAN_CREATE_SCHEMA (Optional) Whether to auto-create the Oban schema at boot. Accepts "true"/"false". Default: true.
• OBAN_KNOWLEDGE_QUEUE_WORKERS (Optional) Number of workers for the knowledge_queue. Default: 1.

Semantic Search Configuration

The following environment variables control the behavior of semantic search:

• SEMANTIC_SEARCH_NUM_CANDIDATES (Optional) Number of candidates to consider in KNN search. Default: 200
• SEMANTIC_SEARCH_K (Optional) Number of top results to return from semantic search. Default: 20
• SEMANTIC_SEARCH_SIMILARITY (Optional) Minimum similarity threshold for search results. Default: 0.30
• SEMANTIC_SEARCH_BOOST (Optional) Boost factor for KNN search results. Default: 1.0

Python configuration

To use and develop knowledges locally, you need a python virtual enviroment.

You need to create a Python virtual environment. This command creates and configures a virtual environment with all dependencies:

mix python.setup

You can specify the python binary to use by setting the PYTHON_BINARY environment variable.

• PYTHON_BINARY (Optional) Absolute path to the python binary. Default: python.
  • Note: This path must be absolute, not relative.
  • The python binary must have the dependencies listed in priv/python/requirements.txt installed.

SSL conection

• DB_SSL: boolean value, to enable TSL config, by default is false.
• DB_SSL_CACERTFILE: path of the certification authority cert file "/path/to/ca.crt".
• DB_SSL_VERSION: available versions are tlsv1.2, tlsv1.3 by default is tlsv1.2.
• DB_SSL_CLIENT_CERT: Path to the client SSL certificate file.
• DB_SSL_CLIENT_KEY: Path to the client SSL private key file.
• DB_SSL_VERIFY: This option specifies whether certificates are to be verified.

AI Provider Proxy Environment Variables

To configure the proxy settings for the AI providers (Gemini and Azure OpenAI), you can set the following environment variables:

• AI_PROVIDER_PROXY_ADDRESS: The address of the proxy server.

• AI_PROVIDER_PROXY_SCHEMA: The schema to use for the proxy connection (default is "http"). This value will be converted to an atom.

• AI_PROVIDER_PROXY_PORT: The port number for the proxy server.

• AI_PROVIDER_PROXY_OPTIONS: Additional options for the proxy configuration.

• KNOWLEDGE_EMBEDDINGS_BATCH_SIZE:

  • Purpose: Controls the batch size used when processing knowledge embeddings in the KnowledgeProcessor worker
  • Default: 100
  • Usage: Defines how many knowledge chunks are processed together in each batch when generating embeddings for knowledge files. Used by the knowledge processing pipeline in td-ai
  • Performance: Adjusting this value can help balance memory usage and processing efficiency when handling large knowledge files with many chunks

Example configuration in your environment:

export AI_PROVIDER_PROXY_ADDRESS="proxy.example.com"
export AI_PROVIDER_PROXY_SCHEMA="https"
export AI_PROVIDER_PROXY_PORT="8080"
export AI_PROVIDER_PROXY_OPTIONS="some_options"