LLMWhisperer is a technology that presents data from complex documents to LLMs in a way that they can best understand.
LLMWhisperer is a technology that presents data from complex documents to LLMs in a way that they can best understand. On Nagent, LLMWhisperer is exposed as a fully-configurable ai document extraction integration that any agent can call — 11 actions, and API key authentication. No code is required to wire LLMWhisperer into your workflow — connect it once via the External Integrations panel and reuse it across every agent you build.
Agent builders use LLMWhisperer to automate the kinds of tasks ai document extraction teams previously handled manually. Concrete examples — each one is a single agent step in Nagent — include:
Every action and trigger is paired with a structured input/output schema (visible in the sections below), so when you wire LLMWhisperer into Helix — our agentic agent builder — the editor knows exactly what each step expects and produces. Configure once, deploy anywhere across your Nagent agents.
Every operation an agent can call against LLMWhisperer, with input parameters and output schema. Drop these into any step of an agent built in Helix.
LLMWHISPERER_CONVERT_DOCUMENT_TO_TEXT_V2Tool to convert PDF/scanned documents to text format for LLM consumption. Supports file upload or URL processing with multiple modes (native_text, low_cost, high_quality, form, table). Use when you need to extract text from documents for LLM processing. Returns whisper_hash for status checking and text retrieval.
Input parameters
Custom tag for usage tracking and auditing purposes.
URL of the document to convert. Provide either url or file, not both.
File to upload and convert. Provide either file or url, not both.
Language hint for OCR. Currently auto-detected but can be specified.
Processing mode for document conversion.
Filename for usage reports and auditing. Auto-populated from file.name if not provided.
Output format mode.
If true, the URL is sent in POST body instead of query param. Only used when url is provided.
Name of registered webhook to call after processing completes. Webhook must be pre-registered.
Add line numbers to extracted text and save line metadata for highlights API.
String to use as page separator in output. Supports dynamic placeholders like '<<< {{page_no}} >>>' to include page numbers.
Specify which pages to extract. Format: '1-5,7,21-' extracts pages 1,2,3,4,5,7,21 to last page.
Custom metadata to pass to webhook callback when processing completes.
Median filter size for noise removal. Only works in low_cost mode. Must be non-negative.
Whether to reproduce vertical lines in document. Not applicable for native_text mode.
Gaussian blur radius for noise removal. Only works in low_cost mode. Must be non-negative.
Whether to reproduce horizontal lines. Requires mark_vertical_lines=true to work.
Line splitter strategy for customizing line splitting behavior in multi-column layouts.
Include line confidence scores in metadata. Requires add_line_nos=true.
Factor to decide when to move text to next line. Default 0.4 means 40% of average character height. Range: 0.0 to 1.0.
Horizontal stretch factor for multi-column layouts. 1.1 = 10% stretch. Must be positive.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_GET_HIGHLIGHTSTool to get line metadata for highlighting extracted text in the original document. Returns bounding box coordinates (x, y, width, height) and page number for each line. Use when you need to create text overlays on document images or highlight specific lines in the source document.
Input parameters
Lines to retrieve metadata for. Format: ranges and individual lines (e.g., '1-5,7,21-' retrieves lines 1,2,3,4,5,7,21 to end). Required if extract_all_lines is false.
The whisper hash returned from the /whisper endpoint. Format: 'hash1\\ | hash2'
If true, extract metadata for all lines. If false, the 'lines' parameter must be provided.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_REGISTER_WEBHOOKTool to register a new webhook endpoint for LLMWhisperer async notifications. Use when you need to set up a callback URL to receive processing results. The webhook URL is validated during registration to ensure it's reachable.
Input parameters
The webhook endpoint URL that will receive callback notifications from LLMWhisperer after document processing is complete. Must be a valid HTTP/HTTPS URL.
Bearer token for authenticating callbacks to your webhook endpoint. Provide the token value without the 'Bearer' prefix. Use an empty string if no authentication is required.
A unique identifier/name for this webhook registration. Must be unique across all your webhooks.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_USAGE_GET_INFOTool to check usage metrics of your LLMWhisperer account. Use when you need to monitor API consumption, verify quotas, or check remaining page limits.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_USAGE_GET_STATSTool to retrieve usage statistics for your LLMWhisperer account based on a specific tag. Use when you need to check consumption metrics for a given tag and optional date range. Returns usage data for the preceding 30 days when date parameters are omitted.
Input parameters
Tag to filter usage data. Required parameter to identify the specific usage metrics to retrieve.
End date for usage period in YYYY-MM-DD format. If omitted, defaults to current date.
Start date for usage period in YYYY-MM-DD format. If omitted, defaults to 30 days before to_date.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WEBHOOK_DELETETool to delete a registered webhook from LLMWhisperer system. Use when you need to remove a webhook that is no longer needed.
Input parameters
The name of the webhook to delete. This is the unique identifier used when the webhook was registered.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WEBHOOK_GET_DETAILSTool to retrieve registered webhook details for LLMWhisperer. Use when you need to get the configuration of a specific webhook including its URL and authentication token.
Input parameters
The name of the webhook to retrieve details for
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WEBHOOK_UPDATETool to update an existing webhook configuration for document conversion callbacks. Use when you need to modify the callback URL, authentication token, or webhook identifier. The system validates the webhook by sending a test payload and requires a 200 status response.
Input parameters
The callback URL that will receive notifications after document conversion completion
Bearer token for authenticating requests to the callback URL. Leave empty if the endpoint requires no authentication
Unique identifier for the webhook configuration
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WHISPER_CHECK_STATUSTool to check the status of a text extraction process in LLMWhisperer. Use when the conversion is done in async mode to poll for completion status.
Input parameters
The whisper hash returned while starting the whisper process. This is used to identify and track the specific conversion job.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WHISPER_GET_DETAILTool to retrieve comprehensive details about ongoing or completed text extraction process. Use when you need to monitor the status and progress metrics of a text extraction job.
Input parameters
Identifier returned when initiating extraction. This is used to retrieve the status and details of a specific extraction job.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
LLMWHISPERER_WHISPER_RETRIEVE_TEXTTool to retrieve extracted text from asynchronous whisper processing. Use when the conversion process was initiated in async mode and you need to retrieve the results using the whisper_hash identifier. Note that retrieval is single-use for security - once retrieved, the same whisper_hash cannot be used again.
Input parameters
When true, returns only the extracted text. When false (default), returns text with metadata including confidence scores and webhook metadata.
Unique identifier returned when initiating the whisper process. Format: 'hash1\\ | hash2'
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
No publicly available marketplace agent is found using this tool yet. There are 58 agents privately built on Nagent that already use LLMWhisperer.
Build on Nagent
Connect LLMWhisperer to any Nagent agent in minutes — no API key management, no boilerplate. Just configure and deploy.
The five questions agent builders ask before adopting a new integration.
Open the External Integrations panel inside Nagent (app.nagent.ai/externalIntegration), find LLMWhisperer, and click "Connect Now." You'll authenticate with an API key — Nagent handles credential storage and refresh automatically. Once connected, LLMWhisperer is available to any agent in your workspace.
No. Nagent provides no-code integration for every tool. Once LLMWhisperer is connected, you configure its 11 actions directly in the agent builder UI — no API calls, no boilerplate, no schema management.
Helix — Nagent's agentic agent builder — lets you drop LLMWhisperer steps into any workflow visually. Pick an action (e.g., one of those listed above), fill in the inputs (Helix knows the required vs. optional schema for each parameter), and connect it to upstream/downstream steps. Triggers run as the entry point of an agent, so when a LLMWhisperer event fires, the agent kicks off automatically.
Every LLMWhisperer action and trigger ships with a fully-typed schema — input parameters with name, type, required flag, and description, plus the output payload shape. The schemas are documented in the sections above. Helix uses these schemas to validate your configuration at build time and to type-check the data flowing between steps.
Yes. While LLMWhisperer ships with 11 pre-built ai document extraction actions, you can layer custom logic around them inside Helix — pre/post-processing steps, conditional branches, retries, or stitching LLMWhisperer together with other connected tools. For deeper customization, talk to our team about Nagent's Agentic AI Lab — forward-deployed engineers who build LLMWhisperer-based workflows tailored to your business.