The SEC’s guidance on internal control over financial reporting says management is responsible for maintaining evidential matter, including documentation, to provide reasonable support for its assessment (Release 33-8810). An AI agent that reads or changes financial data is part of that picture, and the evidence for what it did has to exist, be complete, and be traceable to someone accountable.
ISACA is explicit about agents: every action taken by an AI system should be logged in an audit trail that records who or what initiated it (ISACA, 2025). KPMG warns that agents running on shared service accounts weaken the audit trail, because the ERP log then shows one integration user for every action (KPMG, 2026).
What auditors will ask for
- Every action an agent took in the period, on which system, with what result.
- Who started each run, so a person can be asked about it.
- Evidence the log is complete and retained for the period you need.
- How the listing you hand over was produced, since PCAOB inspectors keep finding untested system-generated reports (PCAOB, 2025).
What engini records
Calls from agents (over MCP, the SDK, the CLI or the REST API) and workflow runs execute through engini, and each is recorded in History under the user who triggered it, even when the connection is a shared integration account. The audit log supports export and streaming to your SIEM, and retention is configurable. Data is encrypted at rest with AES-256 and in transit with TLS 1.2 or later.
engini records the calls that pass through it. It does not record what an agent does by other paths, and it records tool calls, not the model’s reasoning. Keep prompts and conversations in the agent platform if you need them.
NetSuite
SAP S/4HANA
Salesforce