LLMStack is an open-source, no-code multi-agent framework developed by Promptly (trypromptly) that lets individuals and teams build generative AI agents, applications, and chatbots without writing code. Users compose workflows by chaining multiple LLMs together and connecting them to their own data sources and business processes through a visual drag-and-drop builder. The platform supports ingesting data from a wide range of sources including CSV, TXT, PDF, DOCX, PPTX files, as well as Google Drive, Notion, website URLs, sitemaps, and direct uploads. This data can then be used to power retrieval-augmented generation (RAG) pipelines within agents and apps. LLMStack integrates with major model providers including OpenAI, Cohere, Stability AI, Hugging Face, and Google Gemini. It also supports locally-run open-source models via LocalAI, making it suitable for teams with data-privacy or cost constraints who want to avoid cloud APIs entirely. Deployment is flexible: users can self-host on their own infrastructure or use Promptly's managed cloud. The platform is multi-tenant with granular access controls, letting organizations create multiple teams or departments that each see only their own agents and data. Built apps can be exposed via HTTP API or triggered from Slack and Discord integrations. Common use cases include AI SDRs, research analyst bots, RPA automation pipelines, intelligent document processing, and dynamic content generation. The open-source codebase is available at github.com/trypromptly/LLMStack and is actively maintained. Key features: - No-code drag-and-drop builder for chaining LLMs into multi-step workflows - Data ingestion from Google Drive, Notion, PDFs, CSVs, sitemaps, and direct uploads - Multi-provider LLM support: OpenAI, Cohere, Hugging Face, Google Gemini, and local models via LocalAI - Multi-tenant architecture with per-organization access controls - Self-hosted or managed cloud (Promptly) deployment options - Slack and Discord integrations to trigger AI chains - HTTP API access for programmatically invoking built apps - Retrieval-augmented generation (RAG) built into data-connected workflows
Free tier available; Pro tier starts at approximately $50/month. Self-hosted (open-source) is free.