Agent Tools provide various capabilities that an AI assistant or workflow can use. Thinking about these capabilities as families makes it easier to decide what your assistants need without having to understand every individual tool.
A tool is a capability an AI assistant or workflow can use to retrieve information, perform an action, or start a process.
Some capabilities are provided by Applaud. Others are created from connections to your organization's systems.
For information about how Agent Tools works, see Agent Tools overview.
Tool families and tool types
Agent Tools uses two ways to describe tools:
- Tool family describes what a tool helps the assistant do.
- Tool type describes how the tool works.
For example, a tool that checks an employee's leave balance belongs to the Connected systems family and might be an API tool.
A tool that starts an onboarding workflow belongs to the Workflows family and is a Start workflow tool.
This distinction helps you understand both the purpose of a tool and how it's implemented.
Tool families at a glance
| Family | What it does | Typical use |
| Knowledge | Finds information from your organization's knowledge sources. | Answers policy and process questions. |
| Conversation and context | Helps the assistant work with conversation content, files, and relevant context. | Understands an uploaded document or continues a conversation with its existing context. |
| Connected systems | Retrieves live information or performs actions in connected business systems. | Check leave balances, update employee information, or create a ticket. |
| Guided requests | Collects and validates information before completing an action. | Submit a leave request or update an address through conversation. |
| Documents | Generates documents from templates. | Create employment letters or case correspondence. |
| Workflows | Starts or interacts with longer-running processes. | Start onboarding or continue a guided process. |
| Case tools | Performs actions within a Case Management context. | Create tasks, update case plans, or draft case messages. |
| Visuals | Presents information as supported visual content. | Show data in a chart, card, list, or other visualization. |
| MCP | Connects Applaud with external AI clients or external MCP servers. | Let supported AI clients use Applaud or let Applaud use external MCP capabilities. |
Only capabilities that are available and configured for an assistant or workflow can run. An AI assistant can't create or invent a new system action.
Knowledge
What it does: Finds and uses information from your organization's knowledge sources, such as policies, procedures, guides, and supported external content.
What employees experience: An employee asks a policy or process question and receives an answer based on your organization's content.
What admins decide: Which knowledge sources are available and which assistants can use them.
Knowledge helps answer questions such as:
"What's our parental leave policy?"
It doesn't provide live information from business systems. For that, use connected-system tools.
Conversation and context
What they do: Help the assistant work with information and context from a conversation.
Examples include:
- Working with uploaded files
- Finding information in an attachment
- Using relevant conversation history
- Applying supported user context
What employees experience: An employee can upload a document and ask the assistant to find information in it, or continue a conversation without repeating information already provided.
What admins decide: Which assistants and channels can use the available conversation capabilities.
These capabilities are provided by Applaud rather than created from an external system connection.
Connected systems
Connected-system tools let an assistant interact with the applications your organization already uses.
You can connect to systems such as:
- Workday
- SAP SuccessFactors
- Applicant tracking systems
- Learning management systems
- ServiceNow
- Custom APIs
- External MCP servers
A connection provides access to the operations supported by the connected system. You choose which operations to make available as tools.
For example, a connected HR system might provide tools to:
- Check an employee's leave balance
- Find employee information
- Search organizational data
- Update an employee record
How connected-system tools work
You typically:
- Connect the system.
- Discover its available operations.
- Select the operations you need.
- Configure them as tools.
- Test and activate the tools.
- Attach them to the appropriate assistants or workflows.
One connection can support many tools.
You don't need to expose every available operation. Start with the actions your employees and agents actually need.
Guided requests
Guided requests help an assistant collect information across several conversation turns before completing an action.
For example, an assistant might ask an employee for:
- The type of leave they want to request.
- The start and end dates.
- Any additional information required.
- Confirmation before submitting the request.
The assistant then submits the completed request to the connected system.
What employees experience: A short conversation that collects the required information, confirms the request, and completes the action.
What admins decide: Which guided experiences are available, what information they collect, and which assistants can use them.
Guided requests are useful when an action needs more information than the employee is likely to provide in a single message.
Documents
Document tools generate files from templates and available data.
For example, an assistant might generate:
- An employment letter
- A confirmation letter
- Case correspondence
- Other documents your organization has designed
What employees experience: The assistant generates the requested document and provides it as part of the conversation or workflow.
What admins decide: Which templates are available, what data can populate them, and where document generation can be used.
Workflows
Workflow tools connect AI assistants to longer-running business processes.
An assistant can use a workflow tool to start a user-initiated workflow or interact with a workflow that's already in progress, where supported.
For example:
"Start onboarding for this new employee."
The assistant can hand the request to the workflow, where the longer process is managed using the workflow designer.
What employees experience: The assistant starts or continues a guided process without requiring the employee to find and launch the workflow separately.
What admins decide: Which workflows AI assistants can use and how they're exposed to those assistants.
Workflows remain the place to design complex, multi-step business processes. Agent Tools provide the connection between those processes and AI experiences.
Case tools
Case tools are designed to work within a Case Management context.
They can support activities such as:
- Creating and completing tasks
- Building or updating a case plan
- Drafting suggested messages
- Supporting other case-related actions
What employees experience: Case agents and case AI can take structured actions while working on a case.
What admins decide: Which case tools are available to case assistants and how they're used.
Visuals
Visual tools allow an assistant to present information using supported visual formats rather than relying solely on text.
Depending on the channel and tool, this can include:
- Cards
- Lists
- Carousels
- Charts
- Maps
- Organizational charts
- Other supported visualizations
What employees experience: Information is easier to scan and understand when it's presented in an appropriate visual format.
What admins decide: Which assistants can use visual capabilities. The available experience depends on the channel.
A visual tool doesn't change the underlying data. It changes how the assistant presents the result.
MCP
The Model Context Protocol (MCP) lets AI applications discover and use capabilities provided by other systems.
Agent Tools supports two MCP scenarios.
| Direction | What it means |
| Outbound MCP | Applaud connects to an external MCP server and makes its available capabilities usable within Agent Tools. |
| Inbound MCP | A supported external AI client connects to Applaud's MCP server and uses Applaud's assistant. |
These are separate scenarios with different configuration requirements.
Built-in and custom tools
Not every capability is something you create yourself.
Built-in tools
Applaud provides some capabilities as standard platform functionality. These can include knowledge, conversation capabilities, case capabilities, and other standard tools.
You can use these capabilities where they're supported without creating the underlying integration yourself.
Custom tools
You can create custom tools when you need to connect an assistant or workflow to a specific system or business process.
Custom tools can be created from supported connections and capabilities such as APIs, MCP, document generation, guided requests, and workflows.
The available configuration and lifecycle options depend on the type of tool.
Choosing the right tool
When deciding which capabilities to provide, start with what you want the assistant to accomplish.
| If you want the assistant to... | Consider... |
| Answer questions about company policies | Knowledge |
| Understand an uploaded document | Conversation and context |
| Retrieve live employee information | Connected systems |
| Submit a request that needs several pieces of information | Guided requests |
| Create a letter or other file | Documents |
| Start a longer business process | Workflows |
| Take action on an HR case | Case tools |
| Present data visually | Visuals |
| Use capabilities from another AI or MCP system | MCP |
Start with the outcomes employees and agents need rather than publishing every capability a connected system provides. You can add more tools as new use cases emerge.