No-code AI agents that ship with the password 'promptly'

What it does

LLMStack is a browser-based platform for assembling generative AI agents and workflows by chaining multiple LLMs. It ingests documents from sources like Google Drive, Notion, and direct uploads, then handles preprocessing and vectorization automatically. The resulting apps, chatbots, and automations can be exposed via HTTP API or triggered directly from Slack and Discord. The whole system is multi-tenant and can run on-premise or as a managed cloud service.

The interesting bit

The value is in the unglamorous plumbing: LLMStack bundles its own vector database, document preprocessing, and multi-tenant access controls so that chaining models is not the hard part—getting your PDFs and Notion pages into the chain is. That combination of data ingestion, vectorization, and organization-scoped permissions is what moves it from toy demo to internal tool.

Key highlights

• No-code builder for chaining multiple LLMs into agents, workflows, or chatbots
• Built-in data ingestion and vectorization for CSV, PDF, DOCX, PPTX, and TXT from Google Drive, Notion, websites, and uploads
• Multi-tenant architecture with organization-scoped data and user access controls
• Deployable on-premise or used via the hosted Promptly cloud offering
• Integrations with Slack and Discord to trigger AI chains from chat

Caveats

• Ships with a hardcoded default admin password (admin / promptly) that the README explicitly warns you to change after first login
• Background job execution requires Docker; without it, job functionality is unavailable
• Windows installation requires WSL2, limiting native desktop deployment

Verdict

Evaluate it if your team needs a self-hosted, multi-tenant no-code layer for LLM apps tied to internal documents. If you already live in Python notebooks and prefer direct API calls, this will feel like unnecessary middleware.