Features and Capabilities
The C3 Generative AI Application includes built-in capabilities for querying, reasoning, and automating tasks across your enterprise data. This page describes the features available to help you understand how the application supports your work.
The following sections describe the key features:
UI features
Querying and conversation
You can use natural language to explore information across connected data sources. The application understands context, intent, and follow-up questions to support natural, interactive dialogue.
You can perform the following actions:
- Search across internal documents and databases.
- Summarize long files or sections of reports.
- Compare information across multiple datasets.
- Refine or extend answers through follow-up queries.
Recent chat history
The application maintains your recent conversation history so you can review, continue, or reuse previous exchanges. The application retains full context in each chat, allowing follow-up questions to reference earlier responses without restarting the conversation.
Highlight tooltip
The Highlight Tooltip is a popover that appears when a user click-drags to select text (or table/grid content) inside an AI-generated response. A Reply button appears above the selection. Clicking it:
- Places the selected text as a quoted context block above the search input
- Focuses the search bar so the user can type a scoped follow-up question
- Dismisses the tooltip and clears the text selection
- Submits as a new query, with full conversation history preserved
- Automatically clears the quoted context from the search bar after submission
Canvas
Canvas is a dedicated document-creation and editing workspace that activates automatically when an agent produces a long-form output — a draft report, email, form, chart, or any other structured artifact.
In simple terms, whenever you prompt to create a document, Canvas opens as a full-screen editor so you can review, edit, and iterate on the document alongside the agent.
Multimodal inputs
You can include files, images, and text together in a single query. Attach a document, screenshot, or image for additional context, and the application automatically processes all content types in one request.
For example, you can upload a scanned report and ask: "Summarize the performance data shown in this image."
Translation
The application translates automatically across all queries. You can ask questions and receive responses in your preferred language, even when the source content is in another language.
The application grounds all results in your organization's data for traceable and reliable output.
Intelligent agents and tools
Agents
The application uses agents to interpret queries, determine intent, and coordinate reasoning across data and tools. Each agent type has a distinct role and uses specialized toolkits for its tasks.
- Dynamic Agent Canvas: Allows you to work with enterprise data and applications through analysis, document creation, and visualization. Learn more in Canvas Agent.
- Deep Research Agent: Allows you to perform research and documentation, combining in-depth analysis, structured reporting, and content generation. Learn more in Deep Research Agent.
- Workflow Agent: Allows you to automate multi-step processes by coordinating actions between agents, tools, and data systems in sequence.
Agent skills
You can attach reusable skills to agents to give them domain-specific knowledge or behavior for a particular task. Skills define how an agent approaches a problem, what context to consider, and what output format to produce.
For more information, see Agentic Skills Overview.
Agent memory
Agents can store and recall information across conversations using long-term memory. This allows agents to remember preferences, prior decisions, and context from earlier interactions to provide more relevant responses over time.
For more information, see Agent Memories.
The application uses agents to orchestrate tools to perform operations such as searching documents, retrieving structured data, or generating visual outputs.
Tools are part of each agent's toolkit.
Tools
Toolkits group related tools that agents use for search, visualization, document handling, and workflow creation.
Toolkits provide the following capabilities:
| Toolkit name | Associated agent | Description |
|---|---|---|
| Canvas Agent Default Toolkit | Dynamic Agent | Supports data visualization, structured and unstructured search, and document creation or editing. |
| Canvas Agent Deep Research Toolkit | Deep Research Agent | Enables content analysis, summarization, and document replacement across connected sources. |
| Deep Research Default Researcher Toolkit | Deep Research Agent | Adds web search capabilities to extend research beyond internal data sources. |
| Default Agent Toolkit | All agents | Provides core tools for structured search, fuzzy matching, and unstructured query handling. |
| Create Workflow Toolkit | Workflow Agent | Enables workflow creation, graph editing, and node or edge management for automation design. |
The application uses agents to select tools from these toolkits based on the query type and context, combining outputs to deliver accurate, contextual responses.
To learn more about agents and tools, see Agents and Tools Overview.
Workflows
Workflows automate recurring processes that include multiple connected steps. Each workflow links actions such as loading data, analyzing it, and producing results in sequence.
You can combine system actions with agent-driven steps to build repeatable business processes that run automatically without manual effort. Workflows also support document generation, allowing you to produce structured reports and documents as part of automated processes.
To learn more about workflows, see C3 AI Workflows Overview.