BrainPy Playbooks

BrainPy playbooks are guided AI workflows for common accounting review tasks. They help you move from a broad question, such as "what could block close?", into a focused, reviewable packet of findings, context, and next steps.

Playbooks are designed for accountant-in-the-loop work. They can gather context, summarize evidence, and recommend next actions, but they do not replace professional review or automatically change accounting records.

When to use playbooks

Use a playbook when you want a consistent review pattern across clients or periods. Good examples include:

• Checking client collaboration blockers before close.
• Preparing a follow-up packet for outstanding client requests.
• Reviewing AP, AR, cash, inventory, or controls risk.
• Creating a compact evidence summary from the current context.

For one-off questions, the Ask Kraal panel may be enough. For repeatable accounting work, a playbook gives the AI a tighter scope and a more consistent output format.

Available playbooks

Playbook	What it helps with
Client Collaboration Close Blockers	Reviews client requests and advisory signals that could block close work, then prepares a read-only close risk summary.
Client Request Follow-Up	Packages pending, overdue, failed, or unassigned client requests into a concise follow-up packet with next safe actions.
AP Exception Review	Runs a read-only AP stewardship review for a selected client and summarizes evidence-backed AP exception risks.
AR Collection Risk Review	Reviews AR collection risk for a selected client and produces an evidence-linked summary.
Cash Guard Review	Reviews cash protection and bank reconciliation context without executing write actions.
Inventory Variance Review	Reviews inventory variance and item movement risk for a selected client and summarizes evidence-backed findings.
Controls Policy Review	Reviews controls policy posture and client collaboration context without triggering durable changes.
Evidence Bundle Summary	Builds a compact, source-linked evidence summary from the current interactive context. This can be used with or without a specific client when the context supports it.

How to run a playbook

1. Open the relevant client or workspace context.
2. Open Ask Kraal or the AI workflow area.
3. Choose or request the playbook that matches the task.
4. Provide the client, period, and specific concern if the playbook asks for it.
5. Review the resulting summary, evidence references, and recommended next actions.

If a playbook output looks incomplete, ask for a narrower rerun with the missing context. For example: "Run the AP Exception Review for this client for March 2026 and focus on unmatched supplier activity."

Review expectations

Before relying on playbook output, confirm:

• The selected client and period are correct.
• The summary cites or references the source context used.
• Exceptions and recommendations are specific enough to act on.
• Any suggested accounting action is reviewed by the responsible team member.
• The output does not conflict with source-of-record accounting data.

Guardrails

BrainPy playbooks are intentionally bounded:

• They are approved organization workflows, not open-ended background automation.
• They are primarily read-only review and summarization tools.
• They do not perform accounting system write actions by default.
• They should not be used to bypass your firm's approval or review process.
• Organization-level playbook management is restricted to authorized roles.

For sensitive or high-impact accounting changes, use the normal review and approval flow before making changes in your accounting system.

Tips for better results

• Include the client and accounting period in your request.
• Ask for a specific output, such as "close blocker summary" or "follow-up packet."
• Mention whether you want a short executive summary or a detailed review packet.
• Ask the playbook to separate confirmed issues from items that need more evidence.
• Re-run with a narrower scope when the first result is too broad.