> ## Documentation Index
> Fetch the complete documentation index at: https://docs.textmine.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Workbench overview

> Learn how to use TextMine Workbench to ask questions, inspect Vault documents, import files, create reports, run agents, and save reusable artifacts.

Workbench is the AI workspace for getting work done across TextMine. It can answer questions, inspect Vault documents, import files, create reports, run agents, create and run workflows, apply playbooks, update records, fill templates, and save reusable artifacts.

For the full backend capability reference, see [Workbench capabilities](/workbench/capabilities).

<img src="https://mintcdn.com/textmine/Wb04UT80sqhf9k3p/images/intercom/product-updates/26-may-2026/01-screenshot-2026-05-26-at-14-16-18.png?fit=max&auto=format&n=Wb04UT80sqhf9k3p&q=85&s=d6d7c9cd18e379c93e801b7813732da8" alt="TextMine Workbench screen showing an AI workspace for scoped document questions and reusable outputs" width="1705" height="958" data-path="images/intercom/product-updates/26-may-2026/01-screenshot-2026-05-26-at-14-16-18.png" />

## What Workbench can do

* Inspect selected Vault documents before acting.
* Search accessible Vaults and return document or tag tables.
* Keep follow-up work scoped to the selected Vault, folder, document set, uploaded file, provider import, or previous result unless the user asks to broaden it.
* Read Legislate team-document inventory when the user asks about published Legislate documents.
* Ask Vault questions using documents, extracted tags, and uploaded reference files.
* Analyze uploaded files as temporary context without requiring Vault processing.
* Import uploaded files into a Vault when approved.
* Browse and import files from enabled MCP providers such as connected storage systems.
* Create reports, workflows, playbooks, document types, tags, agents, records, and record schemas.
* Run existing workflows, playbooks, agents, table extraction, entity extraction, and document reprocessing.
* Save final answers as Workbench artifacts.

## When to use Workbench

Use Workbench when the answer is not just "open this page and click a button". It is best for multi-step work, exploratory analysis, document comparison, one-off report generation, and drafting reusable automation.

For straightforward document review, start in Vault. For a repeatable process, create a Workflow. For reusable review rules, create a Playbook.

## Workbench and agents

Workbench is the in-app agentic workspace. It is the right place for a human to supervise analysis, review generated outputs, approve Vault changes, and turn one-off work into reusable workflows or playbooks.

External agents should not guess Workbench internals from the UI. They should use:

* [For agents](/getting-started/for-agents) for product routing and terminology.
* [TextMine MCP server](/integrations/textmine-mcp-server) for direct TextMine tool access.
* [API and developer docs](/integrations/api-and-developer-docs) for Public API V3 reference, examples, MCP setup, and exact operation IDs.

When an external agent needs to modify TextMine state, prefer the API or MCP tool that represents the product action directly. Use Workbench when a person needs to collaborate with the model, inspect intermediate reasoning, upload ad hoc files, or approve a change before it becomes durable.

## Common patterns

* Start with Workbench to explore a messy request, then save the repeatable parts as a workflow.
* Upload files into Workbench for temporary context, then import them into Vault only when they need durable processing and governance.
* Name the source scope in the prompt when it matters: a Vault, folder, selected documents, uploaded files, provider import, or previous table.
* Ask Workbench to use Legislate when the question is about validated, published document data.
* Use records when the answer should become structured business data rather than a one-off response.
