Playbooks

Overview

Playbooks are reusable instruction sets that tell the AI how to approach specific types of work. Think of them as expert checklists—when you select a playbook, the AI follows its instructions consistently every time. A “Vendor Agreement Review” playbook might tell the AI to always check for liability caps, indemnification clauses, and unusual payment terms. Playbooks pair with libraries to combine specialized instructions with the right reference documents.

Creating a playbook

1. Click New playbook.
2. Name — Give it a clear, descriptive name (e.g., “NDA Review,” “Patent Analysis,” “Compliance Check”).
3. Description — Briefly explain when to use this playbook. This helps your team know which playbook to pick and helps Arca suggest the right one automatically.
4. Prompt — Write the instructions the AI should follow. This is the core of the playbook—up to 10,000 characters of specific directions for how to analyze, draft, or respond.
5. Libraries — Link one or more libraries that should be used whenever this playbook is active. These libraries are automatically attached when the playbook is selected.
6. Visibility — Choose Private (only you) or Organization (your whole team).

Writing effective prompts

The prompt is the most important part of a playbook. Good prompts are specific and structured. Here are some guidelines:

• Be specific — Instead of “review this contract,” say “identify all liability clauses, flag any uncapped indemnification, and note deviations from our standard terms.”
• Define the output — Tell the AI what format you want: a summary, a checklist, a redlined draft, a risk assessment, etc.
• Include domain knowledge — Mention relevant legal standards, regulatory frameworks, or organizational policies the AI should consider.
• Set priorities — If some issues matter more than others, say so. For example, “focus first on IP assignment clauses, then review termination provisions.”

Using playbooks

Playbooks are applied in two ways:

Manual selection

When starting a chat or working on a request, select a playbook from the dropdown. The playbook's instructions are immediately applied to the AI, and its linked libraries are attached. You can switch playbooks or add additional libraries at any time.

Automatic matching

When a new request is created, Arca analyzes the request's title and description and automatically suggests the best-matching playbook. This means incoming requests from Slack or email can be routed to the right workflow without manual intervention.

You can always override the suggestion and select a different playbook if needed.

Managing playbooks

• Edit — Update the name, description, prompt, linked libraries, or visibility at any time. Changes take effect for future uses immediately.
• Delete — Remove a playbook that's no longer needed. Existing chats that used the playbook are not affected.
• Organization sharing — Set a playbook to organization visibility so your entire team can use it. This is the best way to standardize workflows across your team.

Example playbooks

Here are some common playbook patterns to get you started:

• Vendor Agreement Review — Instructions to identify liability clauses, payment terms, IP provisions, and termination conditions. Linked to a library of your standard vendor templates.
• NDA Review — Instructions to check confidentiality scope, term length, carve-outs, and remedies. Linked to a library of approved NDA forms.
• Compliance Check — Instructions to flag regulatory issues against specific frameworks (GDPR, SOC 2, HIPAA, etc.). Linked to a library of regulatory guidance documents.
• Patent Analysis — Instructions for claim interpretation, prior art assessment, and freedom-to-operate analysis. Linked to relevant patent libraries.

Tips

• Start with a few high-value playbooks for your most common request types, then expand as your team identifies patterns.
• Write clear descriptions so the automatic matching system can suggest the right playbook for incoming requests.
• Link focused, relevant libraries to each playbook rather than attaching everything—more targeted context leads to better AI outputs.
• Share playbooks at the organization level to build institutional knowledge that persists across team changes.