Capabilities
io.Intelligence MCP transforms your io.Connect ecosystem into an AI-ready environment by providing comprehensive capabilities for connecting applications with AI assistants. This page explains what io.Intelligence MCP offers and how it adapts to different use cases.
Capability Overview
io.Intelligence MCP combines discovery, orchestration, controlled exposure, and adaptive behavior in one integration layer.
The approved capabilities below stay the same. This page now simply surfaces them with the same visual rhythm and scanning experience used in the newer module sections.
Capability Areas at a Glance
Built-in system capabilities
Start with application discovery, workspace discovery, launches, orchestration, and optional working context awareness.
Pre-configured functions
Expose stable application capabilities through centralized configuration without changing application code.
Application-managed functions
Let applications decide what functionality is available to AI at runtime and under what conditions.
Adaptive behavior
Adjust discovery, orchestration, and context-sharing behavior based on what the connected AI assistant can do.
The Bridge Between AI and Applications
io.Intelligence MCP acts as an intelligent bridge connecting AI assistants with your business applications. It enables AI to understand what applications are available, what they can do, and how to interact with them - all while maintaining enterprise security and control.
The system automatically adapts based on what the AI assistant can do, ensuring that only appropriate functionality is exposed and that interactions remain safe and controlled.
Application Interaction Methods
io.Intelligence MCP provides three ways for AI assistants to interact with your applications, each designed for different business scenarios:
Built-in System Capabilities
These core capabilities are available immediately and require no additional configuration. They form the foundation of AI-driven workflow orchestration.
| Capability | Description |
|---|---|
| Application Discovery | AI assistants can search through your application catalog to find applications that match user needs. When a user asks to perform a task, the AI automatically identifies which applications can help and explains how they work together. |
| Workspace Discovery | Find and understand pre-configured workspace layouts that match specific business scenarios. The AI can recommend appropriate workspaces based on what the user wants to accomplish. |
| Application Launch | Start applications with the right context and settings. The AI ensures applications open with relevant data already populated, eliminating manual setup steps. |
| Workspace Orchestration | Create new workspaces or restore existing ones with appropriate business context. Launch complete multi-application environments configured for specific workflows. |
| Working Context Awareness | When configured, AI assistants can access the user's current business context - understanding which clients, portfolios, or entities the user is working with. This enables contextually relevant assistance without requiring users to repeatedly specify the same information. |
Pre-configured Application Functions
Expose specific application capabilities to AI assistants through centralized configuration. This approach is ideal for well-defined, stable application features that you want AI to access without modifying existing applications.
Configuration-Based Exposure
Define which application methods and intents should be available to AI assistants in your central configuration. The system automatically manages availability based on whether applications are running.
- Always Available - Functions that are always present and ready to use
- Dynamically Available - Functions that become available when specific applications launch and disappear when they close
Use Cases
This approach works well for established application APIs that need AI integration without application code changes. IT administrators can control AI access to application features through configuration rather than requiring development work.
Application-Managed Functions
Applications can advertise their own AI-accessible capabilities at runtime, giving them full control over what functionality is exposed and when. This approach is ideal for application-specific features that need to adapt to changing conditions.
| Feature | Description |
|---|---|
| Application Control | Individual applications decide when to make their capabilities available to AI assistants. They can expose or hide functionality based on user permissions, application state, or business rules. |
| Flexible Integration | Applications define exactly how AI assistants should interact with them, including what information is needed and what results to expect. This allows for highly specific and contextual AI interactions. |
| Runtime Adaptation | As application conditions change, capabilities can appear and disappear dynamically. For example, an application might only expose certain functions when a user has appropriate permissions or when specific data is loaded. |
This approach is perfect for:
- Applications with context-sensitive functionality
- Features that require special permissions
- Capabilities that change based on application state
Adaptive Behavior Based on AI Assistant Capabilities
io.Intelligence MCP automatically adjusts its behavior based on what the connected AI assistant can do. This ensures safe, appropriate interactions that match the AI's capabilities.
Discovery and Search
When the AI assistant supports elicitation (the ability to request information from users), io.Intelligence MCP enables discovery features:
- Application Search becomes available, allowing the AI to explore your application catalog and recommend the best matches for user requests
- Workspace Search is enabled, letting the AI suggest appropriate workspace layouts for specific business scenarios
This capability is essential for AI assistants that need to understand your application landscape and make intelligent recommendations.
Action and Orchestration
When the AI assistant supports sampling (the ability to invoke AI model interactions for decision-making), io.Intelligence MCP enables action-oriented features:
- Application Launch becomes available, allowing the AI to start applications on behalf of users
- Workspace Orchestration is enabled, letting the AI create or restore complete multi-application environments
This capability ensures that actions requiring user approval are only available to AI assistants that can properly request and handle user confirmations.
Context Management
io.Intelligence MCP adapts its context-sharing behavior based on the AI assistant's built-in capabilities:
- When the AI assistant provides its own context management, io.Intelligence defers to that system
- When the AI assistant relies on external context, io.Intelligence provides working context information
This ensures seamless context sharing without duplication or conflicts.
Business Context Awareness
When enabled, io.Intelligence MCP provides AI assistants with awareness of what users are currently working on. This optional feature dramatically improves the relevance and efficiency of AI assistance.
What is Working Context?
Working context represents the user's current business focus - the specific clients, portfolios, instruments, accounts, or other entities they're actively working with. This information is automatically gathered from applications and shared with AI assistants.
Business Benefits
| Benefit | Description |
|---|---|
| Reduced Repetition | Users don't need to repeatedly tell the AI what they're working on |
| Contextual Intelligence | AI recommendations and actions are based on current work, not generic scenarios |
| Seamless Continuity | When launching applications or creating workspaces, the AI automatically includes relevant context |
| Cross-Application Awareness | Context flows naturally as users work across multiple applications |
Consider enabling working context when:
- Multiple applications share common business entities
- Users frequently switch between applications while working on the same client, account, or portfolio
- You want AI assistance that understands the current business scenario without explicit user input
- Workflow orchestration should respect what the user is currently focused on
Capability Summary
The following table shows how io.Intelligence MCP adapts its available features based on AI assistant capabilities:
| Feature | AI Assistant Requirement | Purpose |
|---|---|---|
| Application Search | Elicitation | Find and recommend applications for user tasks |
| Workspace Search | Elicitation | Discover appropriate workspace layouts |
| Application Launch | Sampling | Start applications with context |
| Workspace Orchestration | Sampling | Create or restore multi-app environments |
| Working Context | None (unless AI handles it internally) | Share user's current business focus |
| Pre-configured Functions | None | Expose configured application capabilities |
| Application-Managed Functions | None | Enable runtime-controlled application features |
MCP Apps
When an AI assistant calls a tool, the result is usually text. MCP Apps extend that model: some tools can also return an interactive HTML panel that the host application renders inline in the chat or as a dedicated workspace window.
MCP Apps follow the MCP Apps standard (mcpui.dev) and can be hosted by any compatible chat application. io.Intelligence adds a custom notification layer on top of the standard that enables response-lock feedback, cross-session state persistence, and programmatic close — all of which are fully open and implementable in third-party chat UIs.
io.Intelligence ships one built-in MCP App as part of the system tools: the Workspace Widget. It lets users visualize an existing workspace layout before opening it, or build a new one interactively through drag-and-drop — all from within the same conversation.
| Capability | Description |
|---|---|
| Inline rendering | App displays inside the chat thread below the tool-call message |
| Workspace rendering | App opens as a standalone window beside the chat in an io.Connect workspace |
| State persistence | Apps save and restore their state automatically across thread switches |
| Response lock | Apps disable interactive controls while the AI is generating a reply |
| Workspace Widget | Built-in system tool for previewing and building io.Connect workspace layouts |
→ See MCP Apps Overview for a full explanation of how MCP Apps work, display modes, lifecycle, and how to enable them in a custom host.
→ See Workspace Widget for the built-in workspace tool, both operating modes, and the custom notification reference.
Next Steps
- MCP Apps - Interactive UI panels returned by tool calls
- Workspace Widget - Built-in workspace visualization and builder
- Transports - Understand deployment options for different environments
- Working Context - Learn about business context management
- API Reference - Technical implementation details for developers