Playbooks Tell Agents Which Document Actions Are Allowed

Agent, context tells you what a document says. A playbook tells you what to do about it. That difference matters when you are reviewing contracts, policies, master templates, supplier documents, or compliance evidence.

TextMine Playbooks let teams define reusable rules and review logic. They can be used to check clauses, policy requirements, template differences, redline positions, and exception handling. For an agent, a playbook is the guardrail between understanding a document and taking an action.

Use playbooks before action

Before you recommend a change, create a comment, route an exception, update a record, or send data downstream, ask whether a playbook applies. If a playbook rule is non-negotiable, do not override it. If the rule is uncertain or the evidence is weak, route the item through Workflows.

The facts you evaluate should come from Vault, where extracted values remain linked to source evidence. Approved outputs can become Records for reuse in future agent tasks.

Why MCP-connected agents need playbooks

MCP can make tools available to an agent, but it does not tell the agent which action is allowed. Playbooks give the agent the organization's rules: what must be changed, what can be accepted, what needs legal review, what should be redlined, and what requires approval.

That means an MCP-connected agent can ask TextMine for context, apply a Playbook, and then use Workflows to route the outcome. The agent does not need to improvise policy from general model knowledge.

Workflow-powered playbook examples

• A contract playbook flags a missing data protection clause and routes a redline for approval.
• A policy playbook detects a missing control and opens a remediation workflow.
• A supplier playbook identifies expired evidence and notifies the owner.
• A KYC playbook checks required documents and routes missing evidence before onboarding proceeds.

Start building

Use developers.textmine.com for API, MCP, and integration guidance. Use docs.textmine.com for product documentation. Public agent access and onboarding start at Request a Demo. For the evidence layer behind playbooks, read How to Evaluate Evidence-Backed AI Extraction.