Dynamic intents, functions or MCP: what to use?

Dynamic intents, functions, and MCPs are three powerful AIsuru tools for extending your Agent's capabilities. Understanding their differences will help you choose the right tool for your specific needs.

Dynamic intents: control and security

Dynamic intents:

• Give you greater control over parameters and conversation flow;

• Don't use generative AI, ensuring predictable responses;

• Are ideal for structured, secure processes.

When to use them:

• To collect specific information (e.g., bookings, forms);

• In situations that require a higher level of accuracy and security.

🔒 Dynamic intents are the best choice when security and predictability are the priority.

Functions: flexibility and power

Functions offer greater flexibility over parameters and use generative AI to interpret results and generate responses.

When to use them:

• For complex, variable interactions (e.g., opening a support ticket);

• For conversations that require depth or creativity.

💡 Functions are ideal when you want to fully leverage generative AI for more natural and adaptive responses.

MCP (Model Context Protocol): intelligent extensions

MCPs (Model Context Protocol) are external servers that extend your Agent's capabilities by connecting it to external services and data sources in real time.

MCPs:

• Let the Agent access up-to-date information and external services;

• Use generative AI to interpret results and weave them into the conversation;

• Are triggered automatically when the Agent identifies a need for them.

When to use them:

• To give the Agent access to real-time data (e.g., web searches, company databases);

• To integrate external services (e.g., CRM, management software, third-party APIs);

• To extend functionality without having to build custom integrations.

🔌 MCPs are the ideal solution when you want to connect your Agent to external systems and enrich its knowledge with always up-to-date data.

Summary table

Here's a table to help you decide between dynamic intents, functions, and MCP. 👇

Feature	Dynamic Intents	Functions	MCP
Control	Greater control, less flexibility	Less control, greater flexibility	Control delegated to external services, maximum extensibility
Generative AI usage	No	Yes	Yes
Accuracy	High, ideal for structured processes	Delegated to generative AI, generally lower in favor of greater flexibility	Depends on the external source; very high for structured and up-to-date data
Response predictability	Consistent and predictable responses	Adaptive and contextual responses	Variable responses based on real-time external data
Complexity	Best for simple, structured processes	Suited to complex, variable interactions	Ideal for integrations with external systems and multiple data sources
Ideal use cases	Collecting specific information Sending emails	Advanced technical support Real-time integration with external systems Creative conversations	Real-time web searches Connecting to company databases Integration with CRM and management software Access to third-party APIs