Skip to main content

Best practices for using Prebuilt Agents

Prebuilt Agents give you a tested starting point for common procedures. Follow these best practices to prepare, customize, and rely on them with confidence in your engagements.

What are Prebuilt Agents?

Prebuilt Agents are custom agents built and tested by DataSnipper, covering common audit and finance procedures. Instead of writing an agent from scratch, you can enable a ready-to-use agent for a procedure you already run, try it, and adapt it to your engagement.

To get the most out of them, and to keep your work defensible, follow the practices below.

1. Treat a Prebuilt Agent as a starting point, not a finished procedure

Prebuilt Agents are designed to be adapted. Before relying on the output, customize the agent to reflect your client's risk assessment and the specifics of your engagement. An agent gets you most of the way there, your judgment makes it fit for purpose.


2. Prepare your workpaper before you run

An agent works with what you give it, so set the workpaper up first:

  • Fill in the header details, such as client name and engagement period.

  • Import all the supporting evidence the procedure needs.

  • Enter any inputs that require your judgment, such as thresholds or tolerances.

  • Add your selected samples to the testing table.

The agent tests the samples and evidence you provide, it doesn't select samples for you.


3. Customize in Your Agents tab and keep the guardrails

Because an enabled agent becomes your own editable copy, you can adjust it in Your Agents just like any custom agent. Focus your edits on what's specific to your engagement, for example, how your client names its documents, or a client-specific rule the agent can't infer on its own.

These guardrails are what keep the output reviewable and defensible, so add to the agent rather than stripping out its built-in logic.


4. Keep judgment and sign-off with the reviewer

Prebuilt Agents document the facts, the conclusions remain yours. Leave reviewer and sign-off fields for a person to complete, and make the final judgment yourself, the agent's job is to do the legwork and present transparent, traceable results, not to sign off.


5. Standardize across your team

Once you've adapted an agent to your methodology, share it so your team runs the procedure the same way every time. If you hold the Content Manager role, you can enable an agent for the whole organization to keep a shared procedure consistent. This is a powerful way for Seniors to turn proven approaches into standard practice and to onboard less experienced team members through guided, agentic workflows.


6. Always review the output

As with any agent, review the results before sign-off. Prebuilt Agents produce transparent, traceable output linked back to the underlying evidence, which makes review faster, but the conclusions and judgment remain yours.

Did this answer your question?