AI Summarizer Component

The AI Summarizer component takes text from one component, generates a summary using an AI model, and writes the result to a second component. Creators write custom instructions to control the tone and format of each summary.

Every AI Summarizer runs on an LLM credential the organization registers. Summaries are generated through the organization's own provider account. Unqork does not supply a model of its own.

The AI Summarizer is a standard component available in the component palette under the Agentic group. Configuration requires wiring an LLM credential, a system prompt, and input and output components. Unqork recommends using the Build Agent to place and configure the AI Summarizer instead of adding it from the palette manually.

Note: The AI Summarizer component will become a Smart Component in an upcoming release.

Use the AI Summarizer component to simplify long documents, condense form submissions, or generate executive summaries for end-users.

General Settings

Setting	Field	Type	Default	Description
Property ID	`key`	Text input	`summarizer`	The unique identifier for this component instance.

Data Settings

The Instructions setting is the system prompt for the summary: it tells the model what to produce and how. The LLM Credential setting selects the provider account that generates the summary, and it determines which models the Model drop-down offers. Model, Thinking Level, Temperature, Thinking Token Budget, and Max Tokens control the quality, length, and reasoning depth of the generated summary. See Understanding AI Model Settings for a plain-language explanation of each.

Setting	Field	Type	Default	Description
Instructions	`instructions`	Text area	—	The system prompt for the summary. Defines the tone, format, focus, and output style — paragraph, bullet points, or a short summary.
	`input`	Drop-down	—	The component providing the text to summarize. Connect to a Text Field or Hidden component, or the output of a Plug-In component.
Output	`output`	Drop-down	—	The component to write the generated summary to. Connect to a Text Field, Hidden, or any display component that accepts a value.
LLM Credential	`llmCredential`	Drop-down	—	Required. The registered LLM credential that generates the summary. An administrator registers credentials under Administration > Environment Settings > LLM Settings > LLM Credentials. Without one selected, the component fails when it runs.
Model	`model`	Drop-down	The credential's default model	The AI model used for summarization. The options come from the provider of the selected LLM Credential — for example, GPT models for an OpenAI credential, or Gemini and Claude models for a Google Vertex AI credential.
Thinking Level	`thinkingLevel`	Drop-down	Low	Controls the depth of the model's internal reasoning before generating a response. Options include: Off, Low, Medium, and High. Applies to models that expose reasoning levels; models that take a reasoning budget read Thinking Token Budget instead.
Temperature	`temperature`	Number	—	Controls response variability. Lower values produce more predictable responses; higher values produce more varied responses. Must be between 0 and 2.
Thinking Token Budget	`budgetTokens`	Number	—	The token budget allocated for the model's reasoning process. Must be between 1 and 4096. Applies only to models that take a reasoning budget instead of a Thinking Level.
Max Output Tokens	`maxTokens`	Number	—	The maximum number of tokens to generate in the summary. Must be between 1 and 65535.

Note: The AI Summarizer component does not use LLM Configs. All model settings are configured directly in the component's settings panel. The Chat Agent Smart Component stores its model settings and system prompt in a server-side LLM Config.

Actions

Setting	Field	Type	Default	Description
Post Trigger	`postTrigger`	Drop-down	—	The Property ID of the component to after the summary generates successfully.
Error Trigger	`errorHandler`	Drop-down	—	The Property ID of the component to trigger if summarization fails.

Execution Context

The AI Summarizer can run in a frontend module (client-side) or an API module (server-side). The right placement depends on when the data to summarize is available and whether system prompt confidentiality matters.

Placement	When to use
Frontend module (client-side)	The input data has not yet been saved to the server — for example, text an end-user is actively typing into a form. The AI Summarizer can read component values before they are submitted.
API module (server-side)	The input data is already saved and can be retrieved server-side. Recommended when the Instructions field (the system prompt) contains sensitive or proprietary guidance that should not be visible to the browser.

When the AI Summarizer runs in a frontend module, the Instructions field is visible in the browser's network tab. Place the component in an API module when system prompt confidentiality is a requirement.

Capabilities

The AI Summarizer component can do the following:

Summarize structured data from component values.
Summarize unstructured text from Text Field or Hidden components.
Process content from files uploaded by end-users, including PDFs.
Apply custom instructions to control tone, format, and focus.
Output summaries as a paragraph, bullet points, or a short summary, depending on the instructions configuration.

Limitations

An LLM credential is required: The AI Summarizer runs on the organization's own provider account. An administrator must register an LLM credential under Administration > Environment Settings > LLM Settings > LLM Credentials, and each AI Summarizer must select one. Unqork supplies no model of its own, so a component saved without a credential fails when it runs.
Models come from the credential: Changing the LLM Credential changes the models available in the Model drop-down. A model the credential's provider does not serve is replaced by that credential's default model.
No guaranteed output format: Summary length and structure depend on prompt engineering and might vary between runs.
No real-time streaming: The summary generates as a complete response, not incrementally.
Token limits: Large inputs might be truncated. PDF files over 10 MB might exceed the model's token limit. Pre-process large documents before summarizing.
Summary quality varies: Accuracy and relevance depend on the clarity of the input and the instructions.

Changelog

Date	Change
2026-08-27	Added component classification (standard/Agentic group), Build Agent configuration recommendation, roadmap note (upcoming Smart Component), and Execution Context section (client-side vs. server-side with security guidance). Removed redundant inline note (EN-8065).
2026-08-20	Added notes on frontend vs. API module system prompt visibility and on AI Summarizer not using LLM Configs.
2026-08-19	Documented the required LLM Credential setting; Model options now come from the selected credential's provider. Corrected the Thinking Token Budget range and its applicability.
—	Initial publication.

Documentation Index