The schedule you already run on: monthly, quarterly, or when the auditor asks.
Evidence work, done before the audit window.
Most of a control test is collection. Pull the report, find the approval, match the ticket, capture the setting, file it with a reference. The judgement at the end takes minutes. The agent does the collection. A person signs.
Control evidence
Highest volumeEach control has a known list of evidence. The agent pulls it from the source systems on schedule and files it against the control, with a record of where it came from.
- Evidence pulled on the schedule each control needs
- Every artefact stored with its source and timestamp
- Gaps flagged weeks before testing starts
Audit requests
Strong fitThe auditor's request list arrives as a spreadsheet. The agent maps each request to evidence already on file and drafts the response pack.
- Requests matched to existing evidence before anyone goes searching
- Missing items assigned to an owner with a due date
Access reviews
Strong fitUser access lists from every system, compared with HR records and role definitions. The agent builds the review. The system owner certifies it.
- Leavers and role changes matched against active accounts
- Exceptions listed for the owner to confirm or revoke
Policy and contract checks
Evidence work onlyReading documents against a requirement. Does this vendor contract have the clause? Does this policy cover the control? The agent finds and cites. A person decides.
- Required clauses located and cited, or marked missing
- Results recorded for the reviewer to accept or reject
The agent collects and cites. A person tests and signs.
Reports, approvals, tickets and settings pulled from the source systems and filed against the control.
Anything missing or out of pattern goes to its owner while there is still time to fix it.
The tester reviews the pack and makes the call. The agent never concludes a control passed.
These are not our numbers. They are the published ones.
We have not published a client result yet. Until we do, here is what KPMG's 2025 survey of SOX programs reports, based on fiscal 2024 data from 146 companies.
SOX Survey, 2025
SOX Survey, 2025
SOX Survey, 2025
SOX Survey, 2025
KPMG's survey covers companies running full SOX programs, so the absolute numbers will be far larger than a mid-size company's. The shape is the useful part. Hours and budgets went up, the share of automated controls went down, and testing is the biggest single block of time.
Which part is worth building first.
Our assessment, not a measurement. The last column decides it. If the work needs a judgement about whether a control works, the agent does not make it.
REPETITION
DEFINED
SYSTEMS
REQUIRED
The agent never concludes that a control passed or failed, and never signs anything. It assembles the evidence and cites every source. A person with the authority to sign makes the call. That line does not move.
Three steps. You can stop after the first one.
We sit with whoever actually clears the exception queue and watch a normal week. Where one of our builds fits, we demo it on your own samples. Either way you get the Unwired Blueprint: one sheet with your process as it runs today, the process unwired, and the line your process owner signs. If the answer is not to build, the sheet says that too. If a scheduled job fixes it cheaper, we tell you that and we are done.
One process, scoped and priced before we start. Built against your systems and your rules, not a demo environment. You see it running on your own data, not on ours.
Code, documentation and runbook handed to your team. No licence, no lock-in, no retainer you have to keep paying to keep it working. Keep us on if you want more built.
Build pricing is set per engagement and quoted in writing before any work starts.
Which control eats the most hours?
Tell us where evidence collection happens by hand. We'll tell you honestly whether an agent is the right fix.