Identity for AI

Tools

Artificial intelligence (AI) agents can extend their capabilities far beyond static training data by leveraging external tools to autonomously solve complex, real-world tasks. Tools are external functions, services, or APIs that an agent can invoke to perform specific actions, retrieve information, or augment reasoning.

Why tools matter

Tools allow AI agents to perform tasks that would be difficult using only their built-in knowledge. With tools, agents can:

Access up-to-date information

Tools enable agents to query external knowledge sources in real time. For example, an agent might:

  • Query a company’s internal database to check a customer’s policy details

  • Access a public API to retrieve the latest stock prices

Perform actions on behalf of users

Agents can interact with systems or services to execute tasks, such as:

  • Scheduling a meeting through a calendar API

  • Sending an email

  • Booking travel arrangements through an external service API

Enhance reasoning and decision-making

Agents can leverage specialized tools to improve accuracy and efficiency, including:

  • Calculators or logic solvers for precise computation

  • Vector databases for semantic search or similarity queries

  • Analytics or simulation engines to evaluate complex scenarios

Integration with protocols

To ensure secure, auditable, and trustworthy tool usage, agents rely on protocols such as Model Context Protocol (MCP) and Agent-to-Agent (A2A).

MCP

MCP provides a standardized framework for agents to discover and invoke tools securely. It enforces identity, authorization, and trust boundaries.

For example, an agent calling a sensitive HR API would require MCP-managed credentials and scopes to perform the action safely.

Learn more in What is Model Context Protocol (MCP)?

A2A

Sometimes an agent delegates a subtask to another agent instead of calling a traditional API. The same security and trust principles apply: the calling agent must authenticate, obtain authorization, and interact according to organizational policies.

Learn more in What is Agent2Agent Protocol (A2A)?

By leveraging tools alongside MCP and A2A capabilities, agents can access information, perform tasks, and collaborate with other agents while maintaining security, user consent, and auditability.

Learn more in A2A and MCP: Complementary Protcols for Agentic Systems in the A2A protocol documentation.