Finance Ethics
Menu

Automate the checks, keep the judgment

We run automated controls, govern AI in finance and build dashboards you own, with people reviewing every flag.

Start a conversationRead about controls for AI agents
Engraving of a brass orrery whose planets are coins orbiting a gear, one coin in magenta.
Systems that keep their own time.

Software assumes staff you lack

Controls monitoring platforms, AI tools and reporting software are built for finance departments with a systems administrator, an internal auditor and time to tune alerts. A company with a controller and two clerks buys the license, connects it, and watches the alerts accumulate because nobody owns them. The tool works. The control does not.

At the same time, AI has reached finance work faster than the controls around it. Invoice-coding agents, reconciliation assistants and general chat tools now handle ledger data in teams that have no inventory of them, no policy for them and no decision about what an agent may approve.

Finance technology at Finance Ethics means we do the work the software assumes someone else will do. We run the monitoring and review every exception before it reaches your team. We write and test the controls around AI. We build dashboards and internal tools that you own outright. Technology is how the work gets done. What you keep is a control that runs.

What we offer here

Each offering has a fixed scope and a named set of deliverables. Most clients start with one and add others later.

  • Engraving of a seismograph drum tracing a steady line with one sharp spike in magenta.

    Continuous Controls Monitoring

    Automated tests on your ledger, bank and access data every week or month, each exception reviewed by a person and routed to a named owner, run as a subscription after a six-week setup.

  • Engraving of a clockwork writing automaton at a ledger, a human hand resting on a magenta brake lever at its back.

    AI Governance for Finance

    An inventory of every AI tool and agent touching finance, a usable AI policy, and controls that treat agents as users, mapped to COSO's 2026 guidance, NIST AI RMF and ISO/IEC 42001.

  • Engraving of a ship's bridge with an arc of brass gauges facing the sea, one needle in magenta.

    Transparency Dashboards & Custom Software

    Board and funder dashboards, accounting data pipelines and custom finance tools that tie to the ledger, live in your own accounts and come with the code and documentation.

Manual checks no longer keep pace

  • AI is already handling ledger data

    Someone drafts journal entries in a chat tool, an AP agent was switched on inside the accounting system, and nobody has listed what these tools can see or do.

  • Reviews happen when someone remembers

    Vendor bank changes, user access and manual journal entries are reviewed after the close if time allows, and the evidence lives in someone's inbox.

  • The board pack is rebuilt by hand

    Every month someone exports the trial balance into spreadsheets, rebuilds the same charts and hopes the totals still tie.

  • An auditor or partner asked about automation

    Your auditor, lender or bank partner wants to know how automated tools and AI agents are controlled, and the honest answer is that they are not, yet.

Automation follows the same four stages

Trace: we map which systems hold the ledger, bank, payroll and access data, which people, service accounts and AI agents act on them, and who owns each control.

Test: we turn the controls that matter into automated tests that run weekly or monthly over full populations, not samples. A person reviews every exception before it reaches your team.

Disclose: confirmed exceptions go to named owners, a monthly exception report goes to management, and a quarterly review shows trends a board can read. Where you publish an Integrity Center, monitoring results support its statements.

Teach: we document every test, threshold and data pipeline in a runbook, train your owners to resolve exceptions, and build everything so you can take it in-house.

Monitoring replaces the annual surprise

An annual audit or a yearly controls review reports what went wrong months after it happened. A vendor's bank details change in February, and the audit sample drawn the following January may or may not include the payment that followed. Continuous monitoring runs the same tests every week or month over every transaction, so a failed control shows up in the cycle in which it failed.

The tests are plain rules wherever possible: a payment released by the person who approved the vendor change, a journal entry posted and approved by one user, an invoice number repeated for the same vendor. Statistical outlier models add a second layer for patterns rules miss, such as a vendor whose monthly spend suddenly doubles. Every flag carries an explanation a controller can read in one line. Our article on ten automated tests lists a typical starting set.

AI agents are users with permissions

When an AI agent codes invoices, matches payments or drafts journal entries, it acts in your accounting system through an account that holds rights. Treat it like any other user. It needs its own credentials, the least access that does the job, and no combination of rights that would be a segregation-of-duties conflict for a person. If it can code an invoice, it should not be able to approve one. Its actions should be logged, and a person should approve anything above a set amount. The agent that approved itself is an illustrative scenario of what happens when that rule is missed.

Three AI frameworks, mapped once

COSO published Achieving Effective Internal Control Over Generative AI on 23 February 2026. It maps the 17 principles of the COSO internal control framework to generative AI, sets out a six-step roadmap that begins with governance and an inventory of AI use, and pays particular attention to shadow AI, meaning tools staff adopt without approval.

The NIST AI Risk Management Framework organizes the same ground into four functions: govern, map, measure and manage. NIST is revising AI RMF 1.0 following the July 2025 AI Action Plan, and its Generative AI Profile, NIST AI 600-1, remains current. ISO/IEC 42001:2023 sets requirements for an AI management system that accredited bodies can certify.

We map one set of finance controls to all three, so you run one program and keep one body of evidence. The AI Governance for Finance offering covers the work in detail.

Built for you to own

Dashboards, data pipelines and internal tools we build run in your accounts, with the code in your repository and documentation your next hire can follow. You own the code and pay no license fee to use it. If we recommend a commercial product instead, we disclose any relationship with its vendor. A monitoring subscription pays for the work of running tests and reviewing results, not for access to your own data.

What people ask before hiring us

Do we have to buy new software?

Usually not. We work from read-only connections or scheduled exports from QuickBooks Online, Xero, NetSuite, Sage Intacct or Microsoft Dynamics 365 Business Central, plus bank files and payroll reports. If a purchase would help, we say which product and why, and we disclose any relationship we have with its vendor.

Who sees the raw alerts?

Our analysts do. Every exception is reviewed by a person before it is routed to your team, so your owners receive items that need a decision rather than a stream of alerts.

Where does our data go?

Into access-logged environments. Client data is never used to train external AI models, and it is deleted 90 days after an engagement closes unless you ask otherwise in writing. Dashboards and tools we build run in your own accounts wherever practical.

Can we take the work in-house later?

Yes. Every test is documented in a runbook and every tool we build is yours, so moving the work in-house is a handover rather than a migration.

Related work and reading

Services

  • Full-Ledger Analytics

    A two-to-four-week diagnostic that runs a library of explainable tests and outlier models over every transaction in the period, then reports only the flags an analyst has reviewed.

  • Segregation of Duties & Approval Design

    A six-to-eight-week project that maps every user's access against their duties, removes conflicting rights, redesigns the approval matrix and adds compensating controls where the team is too small to split.

  • Integrity Center

    A public page on your own domain that shows how your organization governs money, with dated policies, management's control attestations and a change log, kept current by subscription.

Articles and scenarios

Sectors

Tell us what you are seeing

A first conversation is a scoping call: what worries you, what you need to show, and to whom. You leave it with a view on where to start, whether or not you hire us.

Start a conversation
Writehello@financeethics.com
Call(307) 365-5252
Office7124 Commons Dr, Cheyenne, WY 82009
HoursMonday to Friday, 9:00 to 17:00 Mountain Time