Skip to main content

io.Assist Anywhere

ACME Banking has a familiar customer-service problem: their advisors work across several browser applications, each one holding a different part of the client picture. A support agent may need to find the active client, inspect the portfolio, open the right workspace, and ask follow-up questions before they can help.

In this guide, you are the ACME Banking developer adding an AI assistant to that workflow. You will start with a small io.Connect Browser platform, two simple banking apps, two assistant app shells, and a local agentic backend. Chapter by chapter, you will turn that starting point into an assistant-enabled workspace where the AI can understand what the user is working on and help coordinate the available applications.

What You Will Build​

The final application will combine:

  • an io.Connect Browser platform for launching applications and workspaces
  • a client-list app for selecting the active banking client
  • a client-portfolio app for displaying the selected client's portfolio details
  • React and Angular assistant app shells that you can use when building the io.Assist UI
  • an io.Assist experience connected to an io.Intelligence backend
  • a local Mastra backend that serves the assistant agent
  • io.Connect and io.Intelligence APIs that let the assistant understand and act on the workspace

The supporting banking apps are deliberately small. They use plain JavaScript and in-memory data so the guide can focus on io.Assist, io.Intelligence, io.Connect, workspace context, and application interoperability. The assistant app shells are included as clean Angular CLI and Vite React projects so later chapters can focus on adding io.Assist instead of scaffolding framework projects from scratch.

Useful API References​

Keep these references nearby as you work through the guide:

Guide Repositories​

The guide text lives in the io.Intelligence documentation site, but the runnable code lives in the public guide repository:

https://github.com/InteropIO/iointel-guides

Inside that repository, you will use two important directories:

io-assist-anywhere-start/
io-assist-anywhere-solution/

Use io-assist-anywhere-start when following the guide. Use io-assist-anywhere-solution as a reference implementation if you want to compare your work with the completed version.

The local agentic backend lives separately at the repository root:

agentic-backend/

It is intentionally not part of the frontend Lerna package because it is a Mastra backend with its own dependency graph, environment variables, and runtime.

How The Story Unfolds​

ACME Banking starts with a workspace that already contains a client list and a portfolio view. That is useful, but still manual. The guide will progressively add an assistant that can:

  • understand which client the advisor is working with
  • use application and workspace context
  • call application capabilities exposed through io.Connect
  • communicate with an agentic backend
  • become part of a real customer-service workflow

By the end, ACME Banking's support team will have a practical AI assistant embedded in the same workspace where they already do their work.

Next Step​

Before writing code, make sure your machine and licenses are ready in Prerequisites.