runvendo/vendo

Embedded agents your customers use to automate work, build views, and connect their tools.

What it solves

Vendo addresses the problem B2B SaaS teams face when customers request numerous bespoke features. Instead of building every single request manually, Vendo provides a customization layer that allows users to build their own features, views, and micro-apps directly on top of the existing product without requiring changes to the host application's source code.

How it works

Vendo implements an embedded agent that interacts with the product's API as the signed-in user. The process follows three main steps:

  1. Extract: It converts the product's API into tools that the agent can execute.
  2. Generate: The agent creates views and apps using a format-tagged UI document, which are then rendered in a sandboxed iframe to ensure security.
  3. Guard: All actions pass through a guarded tool proxy that manages policies, approvals, and audits to prevent unauthorized access.

Who it’s for

It is designed for B2B SaaS developers who want to give their customers the ability to customize their own UI and automate workflows using plain language, as well as teams looking to expose their product via the Model Context Protocol (MCP).

Highlights

  • User-driven customization: Users can ask for specific views or UI changes (remixing) in plain language.
  • Automated workflows: Enables the creation of standing automations across tools with per-tool approval gates.
  • Sandboxed execution: Generated components run in a secure iframe jail to protect the host application.
  • Flexible integration: Can be installed as a full loop with UI or as a backend-only package for existing agents.

Related

  • Project
    get-bb/bb

    An agentic IDE that can control and customize itself to automate software development workflows through a desktop app, web app, CLI, and HTTP API.

  • Project
    agno-agi/agno

    Agno is a framework and runtime for building and managing agent platforms, providing an SDK, a runtime engine, and a management UI to maintain full control over the agent stack.

  • Project
    github/copilot-sdk

    A set of multi-language SDKs that let developers embed Copilot's agentic workflows and runtime into their own applications.

  • Project
    alan-ai/alan-sdk-web

    An SDK for embedding an intelligent layer into web applications that enables the real-time generation of business logic and UI components.

  • Project
    aws-samples/bedrock-engineer

    An autonomous software development agent app powered by Amazon Bedrock that can create files, execute commands, and manage projects through a customizable agent interface.