The Compliance Risk Hidden in Manual Tolerance Checking
RESPA's tolerance buckets under Reg Z are precise: zero tolerance for origination charges, 10% cumulative tolerance for certain third-party services, and unlimited tolerance for others. A processor manually comparing current fees against the original Loan Estimate in Encompass has to correctly categorize each fee, sum the changes within each bucket, and recognize when a re-disclosure trigger has been crossed — while managing a full pipeline of loans in various stages. Miss a trigger, and you have a RESPA violation. At a non-QM lender with $500M to $3B in annual originations, a 4–6% error rate translates to dozens of violations in any given exam period. The CFPB has been explicit about supervisory expectations for tolerance monitoring.
How an AI Agent Monitors Tolerance Exposure Continuously
An AI Labor Company agent mines Encompass event logs to reconstruct the sequence of fee changes on each loan and understand the tolerance review patterns your processors currently follow. The deployed agent monitors fee-change events in real time, calculates cumulative exposure against each Reg Z tolerance bucket automatically, and identifies re-disclosure triggers the moment they're crossed — not when a processor gets around to checking. Re-disclosure workflows in Encompass are triggered automatically with full documentation of the fee changes that drove the trigger. Every re-disclosure routes to a human approver before it goes to the borrower. Integrations with Black Knight, Reggora, and DocuSign keep the downstream closing workflow intact.
The Business Case: Exam Risk Eliminated, Processor Capacity Recovered
The primary value here is risk avoidance — specifically, eliminating the CFPB exam exposure that comes with a documented error rate in a well-defined compliance process. Moving from a 4–6% RESPA tolerance error rate to near-zero isn't just cleaner operations; it changes the posture you can take in an examination. The secondary value is processor capacity. Removing the manual tolerance-checking burden from each loan in the pipeline frees meaningful time per processor per day — time that goes back to moving loans through the non-QM underwriting process rather than running compliance arithmetic. For a lender at $1B+ in originations, that throughput recovery matters. The agent is typically monitoring live pipelines within about four weeks.
How does the agent handle loans where the original Loan Estimate was issued in a prior system before Encompass?
The agent can be configured to pull baseline LE data from Black Knight or from Encompass import records for loans that originated outside the current LOS. For loans where historical data is incomplete, those are flagged for manual baseline establishment before automated monitoring begins.
Does every re-disclosure require human approval, or can some be automated fully?
Every re-disclosure requires human approval before it is issued to the borrower — this is a deliberate design decision given the regulatory stakes. The agent handles the detection, calculation, and workflow initiation; a human approver confirms each re-disclosure. The agent does not issue disclosures autonomously.
How does this integrate with Optimal Blue for pricing changes that drive fee changes?
Optimal Blue pricing events that result in fee changes on the Loan Estimate are captured as Encompass fee-change events, which the agent monitors. Rate lock changes and repricing events are among the most common re-disclosure triggers in non-QM pipelines and are fully covered by the monitoring logic.