To check today’s weather, KTX schedules, exchange rates, and internal company materials, you need to move between multiple apps and systems. AI that only uses information available at the end of its training has difficulty reflecting continuously changing circumstances.
Tool Calling enables AI to select external tools suited to a user’s request, retrieve the latest information, and execute tasks. This article explains how Tool Calling works, its scope of applications, and what to consider when designing it.
What Is Tool Calling?
Tool Calling is a technology that connects large language models to external functions, APIs, databases, and business systems. After analysing a user’s request, AI determines the necessary tools and input values, then uses the results returned by those tools to provide an answer or perform the next task.
| Category | Training Data-Based Processing | Tool Calling-Based Processing |
|---|---|---|
| Information timing | Uses information available up to the time of model training | Uses external data available at the time of the call |
| Calculations | Generates results based on context | Calls a calculator or calculation function to verify results |
| Internal company materials | Uses only information included in the training process | Queries databases and document systems for which access is authorized |
| Task execution | Generates response text | Performs tasks such as making reservations, registering, sending, and generating reports |
How a Request Is Processed
What You Can Do with Tool Calling
Tool Calling connects information retrieval and task execution in a single conversational flow. Users can provide conditions in natural language without needing to know how to search each service or where its menus are located. Enterprises can configure the tools available for invocation differently according to user permissions and business rules.
| Use cases | Connected systems | Processing examples |
|---|---|---|
| Real-time information | Weather, traffic, foreign exchange rate, and stock price APIs | Retrieve current information and compare it based on conditions |
| Accurate calculations | Calculators, statistical functions, and fare calculation APIs | Calculate taxes, distances, costs, and statistical values |
| Internal information | Document repositories, databases, and search systems | Search regulations, contracts, and performance data within the user's permission scope |
| Schedule management | Calendars and reservation systems | Check available times and add events to the schedule |
| Communication | Email, messaging, and notification systems | Draft and send messages after user approval |
| Business automation | Reporting, customer relationship management, and enterprise resource planning systems | Collect data, complete forms, and change statuses |
Considerations to confirm during design
As connections to external systems increase, the scope of data access and execution permissions also expands. Service operators must define specific verification procedures before and after invocation, error-handling methods, and cost limits. Users must be able to confirm what information is sent to which systems.
| Check item | Potential issue | Design approach |
|---|---|---|
| Privacy protection | Personal information may be sent to external systems during invocation | Send only the necessary data and manage encryption, retention periods, and conditions for providing information to third parties |
| Access permissions | Users may access data or functions they are not authorized to use | Apply user authentication and role-based permissions, and recheck permissions before invocation |
| Data accuracy | An API may return outdated or incorrect values | Display the data timestamp and source, and apply validity checks and cross-validation |
| Execution responsibility | Unintended actions may be confirmed during reservations, payments, or transmissions | Apply user confirmation and approval steps to actions that are difficult to cancel |
| Cost management | Operating costs may increase due to repeated invocations and large-volume requests | Set invocation limits, store results, and establish rules to prevent duplicate requests |
| Maintenance | External APIs and input formats may change | Operate version control, automated testing, failure detection, and fallback procedures |
- Limit sensitive information to the scope necessary for the purpose of the call.
- Require user approval for payment, booking, transmission, and deletion tasks.
- Continuously record the tool’s response time, error rate, and call costs.
- Display the source and lookup time of external data in the results.
- If a tool failure occurs, apply criteria for the number of retries and fallback processing.
CLEVI’s approach to applying AI to real-world work
The usefulness of AI in the workplace depends on model performance, connectable tools, execution accuracy, and the level of security controls. By connecting multiple tools to a single request, you can handle everything from information retrieval to organizing results and carrying out follow-up tasks on the same screen. Reflecting role-specific permissions and approval procedures can reduce repetitive work while maintaining the organization’s security policies.
- Tool Calling
- Function Calling
- Generative AI
- AI agents
- Workflow automation
- API integration