# Introduction > Introduction to io.Intelligence Framework and its package ecosystem # Introduction to io.Intelligence Framework io.Intelligence helps teams make their io.Connect applications AI-ready. Instead of rebuilding existing systems, you can add AI assistants, business context awareness, and standardized tool access on top of the applications and workflows you already have. The framework is designed as a package ecosystem. Some packages help you expose application capabilities to AI systems. Others help you build assistant experiences in the browser or adopt a ready-made assistant offering for your application. ## What io.Intelligence Enables With io.Intelligence, you can: - connect AI assistants to your application ecosystem through Model Context Protocol (MCP) - capture and share business context so assistants can respond more intelligently - let assistants orchestrate tools by writing code that runs in an isolated environment with Code Mode - build custom web assistant experiences on top of io.Connect - add a ready-made assistant UI to an existing application, starting with Angular - bridge a Mastra backend to the io.Intelligence Agent Protocol used by frontend clients - use AI coding agents to learn the framework, scaffold official starters, and accelerate implementation and troubleshooting ## The Package Ecosystem The current documentation covers six module areas that often get combined in real deployments. ### MCP The MCP packages provide the infrastructure layer for exposing tools, resources, and prompts to AI assistants. They cover core abstractions and the main transport options used across the framework. Best fit for: - teams exposing application capabilities to AI assistants - developers building MCP servers and transports - platform teams standardizing AI integration patterns Learn more: [MCP Overview](/modules/mcp/overview) ### Code Mode LLMs are good at writing code. Code Mode uses that strength: instead of calling tools one at a time, the assistant writes a short program that calls the tools it needs and combines their results, and Code Mode runs that program in an isolated sandbox. Applications retain control over which tools the program can reach and what they may do. Best fit for: - tasks that chain several dependent tool calls - workflows that filter, sort, or combine results from multiple tools - assistants working with large or changing tool catalogs - teams that need model-generated logic isolated and sensitive actions controlled Learn more: [Code Mode Overview](/modules/code-mode/overview) ### Working Context Working Context collects and maintains the business context users are actively working with, such as selections, entities, or workflow state. That context can then be shared with AI assistants and applications. Best fit for: - applications that need context-aware behavior - assistants that should understand what the user is working on - teams defining a shared context model across apps Learn more: [Working Context Overview](/modules/working-context/overview) ### AI Web AI Web is the framework-agnostic web SDK for building assistant and copilot experiences on io.Connect. It connects frontend experiences to agent servers, MCP servers, Working Context, and MCP Apps. Best fit for: - teams building a custom assistant UI in the browser - developers who want direct control over assistant behavior and integration - products that need threads, tools, prompts, and resources in one web SDK Learn more: [AI Web Overview](/modules/ai-web/overview) ### io.Assist io.Assist is the ready-made assistant UI offering in io.Intelligence. It is the module area for framework-specific assistant implementations that let teams embed a packaged assistant experience instead of building the full frontend from scratch. The first released item in this area is **io.Assist NG**, the Angular implementation. It is built on top of AI Web and is designed for teams that want to ship an assistant experience quickly inside an Angular application. Best fit for: - teams looking for a packaged assistant experience rather than a custom frontend - Angular teams embedding an assistant into an existing app - projects that prefer configuration over building a custom chat UI - products that want assistant UX, prompt handling, and MCP app support out of the box - teams planning around an offering that can expand into additional framework flavors over time Learn more: [io.Assist](/modules/io-assist-ng/overview) ### AI Server AI Server is the server-side module area in io.Intelligence. It covers backend integrations that supply the route and streaming contract used by frontend assistant clients. The first released item in this area is **AI Mastra Bridge**, which connects Mastra agents to the [io.Intelligence Agent Protocol](/api-reference/ai-server/agent-protocol) used by frontend clients and provides the route layer needed for agent discovery, streaming runs, and thread management. Best fit for: - backend teams evaluating the server-side assistant layer - projects pairing a backend runtime with AI Web or io.Assist implementations such as io.Assist NG - teams using Mastra and looking for the currently released server-side integration Learn more: [AI Server Overview](/modules/ai-server/overview) ## AI-Assisted Development The [io.Intelligence AI Dev Kit](/ai-dev-kit) equips Agent Skills-compatible coding agents with io.Intelligence-specific knowledge and workflows. It can help you choose a package topology, work in an existing project, materialize an official React, Angular, or Vanilla JavaScript application template, follow the io.Assist Anywhere guide, and investigate integration problems. The official documentation remains the source of truth for product capabilities and public APIs. The kit helps an AI coding agent find and apply that information to the actual project in front of it. You don't need an agentic backend to install or use the kit. When you are ready to start an assistant application, its frontend needs a backend that implements the [io.Intelligence Agent Protocol](/api-reference/ai-server/agent-protocol). The kit can help connect an existing backend through [AI Mastra Bridge](/modules/ai-mastra-bridge/overview) or a custom bridge adapter; creating the underlying LLM agentic backend is outside its scope. Learn more: [AI Dev Kit Overview](/ai-dev-kit) ## Choosing the Right Starting Point Start with the package that matches your role: - building infrastructure for assistant access to apps and tools: start with [MCP](/modules/mcp/overview) - letting assistants orchestrate tools through code: start with [Code Mode](/modules/code-mode/overview) - adding business context to assistants or apps: start with [Working Context](/modules/working-context/overview) - building a custom web assistant experience: start with [AI Web](/modules/ai-web/overview) - embedding a ready-made assistant experience: start with [io.Assist](/modules/io-assist-ng/overview) - evaluating the backend side of the assistant stack: start with [AI Server](/modules/ai-server/overview) - using an AI coding agent to learn, scaffold, integrate, or troubleshoot: start with the [AI Dev Kit](/ai-dev-kit) ## Common Package Paths Many integrations use more than one package: - infrastructure-first path: start with [MCP](/modules/mcp/overview), then add [Working Context](/modules/working-context/overview) if assistants need live business context - custom frontend path: pair [AI Web](/modules/ai-web/overview) with [MCP](/modules/mcp/overview), optionally adding [Working Context](/modules/working-context/overview) - ready-made assistant path: start with [io.Assist](/modules/io-assist-ng/overview) for the current Angular implementation, which builds on [AI Web](/modules/ai-web/overview), and connect it to a backend that implements the [io.Intelligence Agent Protocol](/api-reference/ai-server/agent-protocol), such as [AI Mastra Bridge](/modules/ai-mastra-bridge/overview) - Mastra backend path: start with [AI Server](/modules/ai-server/overview), then continue to [AI Mastra Bridge](/modules/ai-mastra-bridge/overview) and point [AI Web](/modules/ai-web/overview) or [io.Assist NG](/modules/io-assist-ng/overview) at its routes ## Getting Started :::warning[Prerequisites] The exact prerequisites depend on the package you use, but common requirements include: - an active **io.Intelligence license** - an **io.Connect Desktop** or **io.Connect Browser** environment - registered applications in your io.Connect ecosystem - a backend that implements the [io.Intelligence Agent Protocol](/api-reference/ai-server/agent-protocol) when you run a frontend assistant experience ::: ### Next Steps If you are evaluating the framework, start with the modules section: - [MCP Overview](/modules/mcp/overview) - [Code Mode Overview](/modules/code-mode/overview) - [Working Context Overview](/modules/working-context/overview) - [AI Web Overview](/modules/ai-web/overview) - [io.Assist](/modules/io-assist-ng/overview) - [AI Server Overview](/modules/ai-server/overview) If you want an AI coding agent to help you evaluate or build with the framework, start with the [AI Dev Kit](/ai-dev-kit) and its [installation guide](/ai-dev-kit/getting-started). If you are ready to integrate, move into the API reference: - [MCP Core API Reference](/api-reference/mcp-core/overview) - [Code Mode API Reference](/api-reference/code-mode/overview) - [MCP Web API Reference](/api-reference/mcp-web/overview) - [MCP HTTP API Reference](/api-reference/mcp-http/overview) - [Working Context API Reference](/api-reference/working-context/overview) - [AI Web API Reference](/api-reference/ai-web/overview) - [io.Assist NG API Reference](/api-reference/io-assist-ng/overview) - [Agent Protocol Reference](/api-reference/ai-server/agent-protocol) - [AI Mastra Bridge API Reference](/api-reference/ai-mastra-bridge/overview)