LLMStack

What it is

LLMStack is a no-code platform for building generative AI agents, workflows and chatbots, connecting them to your data and business processes.

How it works

Features include agents, chaining multiple LLMs, data import with preprocessing and vectorization, a no-code builder for AI chains, and deployment to cloud or on-premise. Apps can be accessed via HTTP API and can be triggered from Slack or Discord. It supports multi-tenant organization structures and keys management for providers like OpenAI. Documentation and quickstart resources are linked in the README.

Getting started

Installation command:

pip install llmstack

Then start the service with:

llmstack

Additional notes:

• Prerequisites mention a background docker container for jobs; Docker installation guidance is linked.
• The default admin account credentials are admin and promptly, with a reminder to change the password after logging in.
• On first run, it creates a .llmstack directory in the home folder and stores database/config files there; browser opens to http://localhost:3000.
• You can add provider keys from the Settings page or via ~/.llmstack/config for default keys.

Recent releases

• v0.2.6 (2024-11-05): Voice agents with OpenAI realtime API; Twilio voice integration; Bug fixes
• v0.2.5 (2024-09-25): Sheets launch; AI sheet builder; Bug fixes
• v0.2.4 (2024-09-13): Sheets; Token usage info in history; Message feedback support
• v0.2.3 (2024-08-14): Bug fixes
• v0.2.2 (2024-08-06): Add new default apps; Allow using file converter

Traction

Stars: 2309; Forks: 347; Open issues: 24

Behind the repo

Website links point to Promptly for cloud offering and documentation; repository indicates a Python-based no-code platform with a Docker-dependent backend.

Caveats

License is not listed in the repo metadata. Creation date: 2023-08-06; last push: 2024-12-11. There are 24 open issues reported.