Agent Knowledge is Fieldguide’s proprietary context retrieval layer that grounds Field Auditor in your firm’s institutional knowledge: from auditor-verified, prior-year workpapers, to your firm’s proprietary methodology and procedures, to regulatory guidance from bodies, so every test reflects your firm’s proven patterns and voice.
*What is Field Auditor? Field Auditor is Fieldguide’s family of AI agents purpose-built for audit and assurance work. Testing Agent is the member of the Field Auditor family that tests controls on your engagements. Agent Knowledge powers Testing Agent by giving it access to your firm’s context, so when it tests a control, it draws on how your firm has approached similar work before.
How It Works
Agent Knowledge runs in three steps:
Ingest: Fieldguide indexes your firm’s previously-tested controls by client, with each client’s knowledge base fully isolated and never shared across clients. Beyond prior engagements, Agent Knowledge will also incorporate your firm’s context including methodology, procedures, testing behavior of your auditors and relevant regulatory guidance, building a comprehensive context graph unique to your firm.
Retrieve: When Fieldguide’s Testing Agent is about to test a control, it searches your firm’s knowledge base for the most similar controls your firm has tested before.
Boost: Testing Agent uses those examples as reference when testing the current control, the same way an auditor would reference a prior-year engagement when testing current-year work.
Every completed engagement adds new auditor-verified examples to your knowledge base, so the Testing Agent draws from a richer set of prior work each time it tests a control. When your reviewers accept or correct an agent’s output, that feedback further refines future runs, thereby creating a self-improving cycle. The more your team uses Fieldguide, the more precisely the agent matches your firm’s rigor, voice, and judgment. With Agent Knowledge, Testing Agents save you an estimated 65% of control testing time.
Security and Privacy
Your firm’s data is never used to train or fine-tune any AI model. Your knowledge base is fully isolated to your firm: no data crosses between firms. Every example Testing Agent references is logged with the user, timestamp, and ordering, so any agent run can be fully reconstructed. Your workpapers are never rewritten or summarized, and your data stays in your firm’s region with the same encryption and access controls as your engagements.
Frequently Asked Questions
Which frameworks are supported?
Agent Knowledge supports all frameworks.
Which controls are included?
Only controls from completed engagements that have linked evidence and documented conclusions. Draft work, incomplete engagements, and controls without supporting documentation are excluded.
Will the Testing Agent still flag real exceptions?
Yes. Agent Knowledge calibrates the Testing Agent’s judgment; it does not lower the testing bar. The improvement is in reducing unnecessary flags, not in suppressing valid ones.
How does Agent Knowledge improve accuracy?
By showing the Testing Agent how your firm has historically tested similar controls, Agent Knowledge calibrates the agent’s judgment to match your firm’s standards. This reduces false exception flags and produces documentation that is closer to your firm’s voice and level of detail.
Does the knowledge base update automatically as we complete more engagements?
Yes. As your firm completes engagements and documents conclusions, those controls become eligible for the knowledge base. Additionally, reviewer corrections and accepted outputs feed back into the system, so the agent’s judgment sharpens with every interaction.
Is my firm’s data used to train AI models?
No. Agent Knowledge is context used at test time only. Your data is never used to train or fine-tune any model: Fieldguide’s or anyone else’s.
Where is my Agent Knowledge data stored?
Your Agent Knowledge data stays in your firm’s region and is protected by the same encryption, backup, and access controls as the engagements it was built from.
What kinds of context does Agent Knowledge include?
Agent Knowledge draws primarily from your firm’s completed prior-year engagements. Over time, the knowledge base will expand to incorporate your firm’s proprietary methodology and procedures, regulatory guidance from bodies like the AICPA and PCAOB, and insights from your team’s corrections and feedback on agent output, building a comprehensive, firm-specific context graph.