Epimetheus is a Slack bot that integrates with local Language Models (LLMs), keeping AI interactions within the organization. It uses Slack Emojis to allow users to choose between different models, such as code-based or image-generating models.
- Model selection: Users can select from multiple LLMs using Slack Emojis, allowing them to choose the most suitable model for their task.
- Customized chat bot: Epimetheus will soon incorporate AI tools and RAGs (Reusable Aggregations) to provide a fully customized chat experience for users.
- Custom prompts: Per-model custom prompts based on the selected Emoji.
Currently it is a bit of an involved installation. The requirements are a Redis instance, an Ollama instance, a Slack App and permissions, and this app itself.
Included is a epimetheus.kube.yaml file and a Containerfile. It is set to run on Podman 🦭, but can easily be
run on Docker 🐳.
Build the Epimetheus Application using podman build . -t epimetheus-app:latest or the docker command.
Create a secrets file in the following format:
apiVersion: v1
data:
slack-app-token: base64-app-token
slack-signing-secret: base64-signing-secret
slack-bot-token: base64-bot-token
kind: Secret
metadata:
creationTimestamp: null
name: slack-appThen you can run the application:
podman kube play epimetheus.secrets.yaml
podman kube play epimetheus.kube.yamlFuture features include:
- RAG support: Local private documents supported in a RAG (Reusable Aggregation).
- Image generation: Using Comfy UI to generate images based on user input.
