Promptise Foundry
A full-stack framework for building, running and shipping AI agents, bundling tool discovery, memory, security, multi-tenancy, human approvals and observability behind a single Python decorator.
Promptise Foundry is a framework meant to replace the separate model SDK, tool layer, vector database, auth, guardrails, job runner and logging setup that agent projects otherwise assemble by hand. A build_agent() call and a Python decorator define an agent and its tools, and capabilities such as memory, a security scanner for prompt injection and PII, response caching, sandboxed code execution and tracing are each switched on with a single parameter.
An agent finds its tools on an MCP server on its own, and Foundry works with OpenAI, Anthropic, Gemini, a local Ollama model, or any model built on LangChain. It installs with pip, and the same install that runs one agent on a laptop is intended to run a multi-tenant fleet serving production users.