TextMine Help Center

Workflow blocks are the building units of a TextMine workflow. A workflow starts from a trigger, can route through one or more conditions, and then runs action blocks in order. Use this page to understand product behavior. For exact JSON schemas, API operation IDs, and examples, use the developer hub’s Public API reference and workflow catalog endpoints.

How blocks work

Each block has a type, configuration, input context, and output. Later blocks can use earlier outputs through labels or variables.

Triggers decide when the workflow starts.
Conditions decide which route should run.
Actions do work, create outputs, or change TextMine or external systems.

Blocks that create durable effects should be treated carefully by agents. Request confirmation before publishing, sending email, changing assignments, creating records, moving documents, submitting to integrations, or triggering a workflow with business side effects.

Trigger blocks

Block	What it does	Use it when
`document_imported`	Starts as soon as a document is added to a Vault.	New documents should immediately enter an automation.
`document_processed`	Starts when a Vault document finishes processing.	The workflow needs extracted text, tags, or processed document state.
`document_published`	Starts when a Vault document reaches the published state.	Publication should trigger follow-up work, reporting, or Legislate handoff.
`email_received`	Starts when an email arrives at the workflow’s inbound address.	Email should create documents, trigger review, or route attachments.
`intake_form`	Creates a public intake form and starts on submission.	External users should submit structured data or files into TextMine.
`scheduled`	Runs on a recurring schedule against selected documents.	The process is periodic, such as weekly monitoring or monthly reporting.
`manual`	Runs only when explicitly triggered.	A human or API client should decide when the workflow runs.

Common trigger configuration includes Vault scope, sender access rules, form fields, upload handling, schedule time, timezone, and document selection rules.

Condition blocks

Block	What it does	Use it when
`confidence`	Compares a document confidence score with a threshold.	Low-confidence extraction or review should route differently.
`document_type_is`	Checks whether the triggering document has a specific document type.	One workflow handles several document types but needs different routes.
`entity_contains`	Checks whether a named entity contains a configured value.	A clause, party, jurisdiction, or extracted field should drive routing.
`ai_logic`	Evaluates a natural-language rule and returns a true or false route.	The routing logic needs semantic judgment rather than a fixed field check.
`decision_policy_matrix`	Evaluates ordered policy rules and selects a decision route.	The workflow should follow a business policy, matrix, or threshold table.

Conditions should be as specific as possible. Use deterministic conditions for known fields and AI logic only when the condition needs interpretation.

Human routing and review actions

Block	What it does	Notes
`add_reviewer`	Adds users as reviewers on the matched document.	Use for human review queues.
`add_assignee`	Adds users as assignees on the matched document.	Use when work ownership should change.
`create_alert`	Creates a task, reminder, or alert.	Can include recipients and due dates.
`request_approval`	Pauses the workflow until approvers approve or reject.	Use before publishing, sending, external submission, or other high-impact actions.

AI, Workbench, and playbook actions

Block	What it does	Notes
`ai_review`	Runs an AI review pass against extracted tags using an instruction.	Best for validating extraction or review quality.
`run_general_action`	Runs an analytical instruction and captures a labeled output.	Use for summaries, classifications, or reusable reasoning outputs.
`run_workbench_capability`	Runs Workbench from a natural-language instruction using the workflow’s Vault, document, and input scope.	Use when the workflow needs Workbench-style planning or analysis as a step.
`run_agent`	Triggers a preconfigured agent and waits for the result.	The agent result becomes workflow output.
`apply_playbook`	Runs a playbook against documents and can create a Word artifact.	Use when review rules should be reusable and auditable.

Values and data pipeline actions

Block	What it does	Notes
`set_values`	Seeds static labeled values for downstream steps.	Useful for controlled lists, constants, or manual inputs.
`calculate_values`	Applies formulas, variables, and thresholds to workflow inputs.	Use for numeric or rule-based calculations.
`organise_values`	Extracts, normalizes, deduplicates, or reformats values.	Use before records, reports, or API calls.
`reconcile_data`	Compares inputs or outputs and reports breaks or exceptions.	Use for matching, variance checks, and exception reports.
`compare_to_accepted_output`	Compares generated output with an approved answer, table, CSV, JSON, or Excel file.	Use for QA and regression-style checks.
`run_python`	Runs a bounded Python data-analysis script over workflow inputs, labels, arrays, variables, and table outputs.	Use for calculations, table cleanup, exception summaries, and intermediate values. REST and network calls are blocked; use `api_request` for external APIs.
`enrich_values`	Searches web or named sources and can download files.	Use when external context is needed.
`api_request`	Runs a structured REST API request.	cURL or Python snippets can be imported as configuration, but arbitrary code is not executed. Responses can be captured as text, JSON values, or downloadable files.
`webhook`	Calls an outbound webhook or HTTP callback.	Uses the same structured request model as API request and can use secret placeholders in headers or bodies.
`tableau_export`	Exports Tableau views or workbooks.	Supports CSV, Excel, PDF, PNG, and PowerPoint outputs where the Tableau integration is configured.
`power_bi_export`	Exports Power BI reports.	Supports PDF, PowerPoint, and PNG outputs where the Power BI integration is configured.

run_python exposes workflow context such as workflow_input, collection, inputs, arrays, variables, tables, workflow_tables, vault_tables, agent_tables, workbench_tables, and tables_by_label. Workflow labels are also exposed as snake_case variables with *_list arrays. Top-level result keys and public variables can be saved as workflow variables for later blocks.

Records and analytics actions

Block	What it does	Notes
`lookup_record`	Searches business records by query or name.	Saves matches under an output label.
`create_or_update_record`	Creates or updates a business record and can link workflow documents.	Use when workflow output should become structured business data.
`get_analytics_metric`	Reads an analytics metric for a scope and period.	Current catalog includes pages processed.
`sample_population`	Selects a fixed or percentage sample and emits a sample register.	Use for audit, QA, and population testing.

Document, source, and ingest actions

Block	What it does	Notes
`workbench_uploaded_files`	Uses files attached in Workbench as workflow input.	Good for turning supervised Workbench work into automation.
`upload_to_vault`	Uploads inbound attachments into the workflow Vault.	Common after email triggers.
`move_document_to_folder`	Moves the current workflow document scope into a target Vault folder.	Use after intake, classification, approval, or exception routing.
`extract_text`	Extracts plain text from submitted files.	Common after intake form upload.
`resolve_ecfr_sources`	Extracts CFR/eCFR citations, fetches official eCFR text, and can follow capped one-hop cross-references.	Use before regulatory analysis, comparison, or report steps.
`resolve_fca_handbook_sources`	Extracts FCA Handbook references, fetches official provision text, and can follow capped one-hop cross-references.	Use when FCA source text should ground downstream analysis.
`search_edgar_filings`	Searches SEC EDGAR filing text and saves matching filings.	Use when filings should feed later workflow steps.
`fetch_sec_filings`	Imports recent SEC filings for companies in scope.	Supports filing type and folder behavior configuration.
`wait_for_document_processing`	Blocks until in-scope documents finish processing.	Use before steps that need extracted text, tags, or document state.
`extract_tables_or_line_items`	Extracts tables or line items from documents.	Can target specific table questions or extract detected line items.

Report and template actions

Block	What it does	Notes
`fill_template`	Fills a spreadsheet, CSV, text, Word, or form template from workflow data.	Can use AI field matching.
`send_report_template`	Generates a report from a predefined template.	Use when a standard report exists; workflow document context is passed into the report step.
`send_report_custom`	Generates a report from a free-form prompt.	Use for one-off or flexible report outputs grounded in the current workflow scope.

Delivery and integration actions

Block	What it does	Notes
`send_to_legislate`	Publishes in-scope imported documents to Legislate.	Use after validation when documents or data should enter the Legislate archive.
`import_from_integration`	Pulls records and attachments from an external system into the workflow Vault.	Current catalog supports ServiceNow.
`submit_to_integration`	Pushes outputs to an external system.	Supports SharePoint file uploads and ServiceNow submissions for attaching workflow artifacts, updating records, or both where enabled.
`send_email`	Sends an email notification and can attach outputs, reports, documents, templates, or intake attachments.	Use sender modes such as original sender or intake submitter for inbound workflows.

Agent guidance

Agents should read workflow catalogs before creating or editing workflow definitions. The TextMine MCP server exposes catalog tools for triggers, conditions, and actions, and the Public API exposes the same catalog for exact parameter schemas. When designing a workflow, agents should return:

Trigger kind and scope.
Conditions and routing assumptions.
Action sequence and expected outputs.
Side effects that need human confirmation.
IDs needed for audit, such as workflow ID, run ID, Vault ID, document IDs, record IDs, and report IDs.

Start here

Use Workbench

Manage documents and data

Automate work

Connect TextMine

Admin and updates

Workflow blocks

How blocks work

Trigger blocks

Condition blocks

Human routing and review actions

AI, Workbench, and playbook actions

Values and data pipeline actions

Records and analytics actions

Document, source, and ingest actions

Report and template actions

Delivery and integration actions

Agent guidance

​How blocks work

​Trigger blocks

​Condition blocks

​Human routing and review actions

​AI, Workbench, and playbook actions

​Values and data pipeline actions

​Records and analytics actions

​Document, source, and ingest actions

​Report and template actions

​Delivery and integration actions

​Agent guidance

How blocks work

Trigger blocks

Condition blocks

Human routing and review actions

AI, Workbench, and playbook actions

Values and data pipeline actions

Records and analytics actions

Document, source, and ingest actions

Report and template actions

Delivery and integration actions

Agent guidance