Skip to main content

Agent Builder - How to create and use custom Agents with DataSnipper Agent Builder

Build and run a reusable custom agent in DataSnipper Agent Builder, with a loan amortization testing example covering goals, behavior, instructions, and execution in Excel. Requires v26.1 or later.

Learn how to turn your audit and finance workflows into reusable, custom agents with DataSnipper Agent Builder.

What is Agent Builder?

Agent Builder turns your audit and finance workflows into custom agents your whole team can run. Instead of starting from scratch every time, you describe a procedure once, including the steps, logic, and decisions behind it, and Agent Builder captures it as a reusable agent that executes the same way every time.

Once created, agents can be shared across engagements and used directly in Excel through Excel Agents. The result is consistent, high-quality work with explainable, traceable outputs, so your best way of working becomes available to everyone.

Agent Builder is not a prompt library. Each agent is a full custom workflow with structured capabilities and guardrails, allowing it to execute complete procedures end to end.

Prerequisites

  • DataSnipper v26.1 or later

  • Accelerate or Elevate DataSnipper package

  • Access to the internet

Accessing Agent Builder

1. Go to app.datasnipper.com in your browser.

2. Enter your email address and submit.

3. Enter the verification code sent to your email.

4. Select your organisation to sign in.

5. Open Agents from the left-hand navigation to start building. Publishing and sharing agents firm-wide depends on your assigned role.

Step-by-step tutorial

1. On the DataSnipper platform, open Agents and click Build an Agent.

2. Give the agent a clear name and short description, for example: "Loan Amortization Testing Agent. Performs loan amortization testing as part of your financial statement audit."

3. In Goals, state the outcome the agent should achieve, for example: re-perform amortisation calculations and confirm the recalculated interest expense agrees to the trial balance.

4. In Behavior, set the guardrails the agent should follow, for example: extract key terms from loan agreements before calculating, and flag any loan where the interest rate or day convention is missing.

5. In Instructions, list the steps to execute in order: extract key terms (capital amount, initiation date, interest terms, day convention, repayment terms, signatures) into the summary table, recalculate the amortisation schedule, tie actual repayments to the bank statements, and compare the recalculated interest expense to the trial balance.

6. Change Status to Published and set Available to (for example, Organization) so your team can use the agent.

7. Open your workpaper in Excel, import the supporting documents into the DataSnipper Document Viewer, and click the Excel Agent icon in the DataSnipper ribbon.

8. Select your agent from the panel and prompt it to execute, for example: "Execute the agent."

9. The agent reads the workpaper, creates snips, executes the defined steps, and writes the output back to your Excel sheets.

10. Review the output, check the snips and formulas for traceability, and apply your professional judgement before sign-off.

Did this answer your question?