The Four Cash Questions No Owner Can Answer Without Opening Something (and the AI CFO Agent That Answers Them)

The Four Cash Questions No Owner Can Answer Without Opening Something (and the AI CFO Agent That Answers Them)

Shawn Kercher
September 15, 2026
AI CFOFinance AutomationCash FlowAI AgentsSmall BusinessThe Brief

The Four Cash Questions No Owner Can Answer Without Opening Something (and the AI CFO Agent That Answers Them)

Try this before you read the rest. What's coming in this week, what's going out, what's owed to you, and what you owe, down to the vendor. Could you answer that right now, out loud, without opening a single thing?

I asked a room full of Denver owners this exact question at Colorado Startup Week and had them raise a hand. Almost nobody did. Mine would not have gone up either a year ago.

Most businesses run on a monthly report and gut

Most businesses run on a monthly report from the bookkeeper and gut for everything in between. Do we hire? Can we afford this? Should we take that job? Every decision between reports is a guess, and the gap between reports is where the money leaks. Not in one bad decision. In thirty small ones made on a feeling.

How I run it now

I have an agent that acts like a CFO. It is connected to my bank, my Stripe, my QuickBooks. Every morning before I start, I ask it what is coming in, where the money is going, and what I can green-light. It answers from the real accounts, not from a report someone built two weeks ago. "Can we afford to hire" stopped being a feeling and started being an answer.

That is not a tech thing. It is a connected-systems thing. Your bank, your books and your invoices already hold every number you need. They just do not talk to each other, and you are the one doing the translating, usually late and usually from memory. Put one agent on top of the three and the translating stops.

Nothing got replaced to do that. Same bank. Same books. Same invoicing. One layer on top that can answer a question.

What the agent actually does

For anyone who wants the mechanics rather than the feeling:

  • Reads, does not write. It has read access to the bank feed, the payment processor and the accounting system. It cannot move money or post entries. That is a deliberate boundary, and it is the reason the security review took an afternoon.
  • Reconciles before it answers. "What's owed to you" is open invoices from the books, netted against payments the processor shows as settled but the books have not caught up with yet. The number it gives me is the real one, not the one from either system alone.
  • Flags the mismatch. When a vendor invoice does not match the purchase order or the amount already paid this year, it says so. At the Startup Week demo it caught one live, on screen, and answered "how much have we paid this vendor this year" with a real dollar figure.
  • Runs in our own account. The data stays in our cloud, under our logins. The own-cloud deployment guide describes the same architecture for a client.

The four questions: do this today

Do this today, and do not guess. Write down your own answer to each:

  1. What's coming in this week
  2. What's going out this week
  3. What's owed to you
  4. What you owe

Then go find the real numbers and write those next to your answers.

The size of the gap is the point. If you were close on all four, you have a system already, even if it lives in your head, and the win is getting it out of your head. If you were off on two or more, you are running on a feeling, and every decision you made last month was made on that feeling. Either way you now know exactly which question you would want answered every morning. That is your first project.

Get the four questions answered every morning

A discovery session maps your bank, books and invoicing, confirms read-only access, and scopes a CFO agent that runs in your own account.

Scope a CFO agent

Frequently asked questions

What is an AI CFO agent?

An agent with read-only access to your bank, payment processor and accounting system that answers cash questions from the live accounts: what is coming in, going out, owed to you and owed by you. It reconciles across the three systems rather than reading one report.

Does an AI CFO agent replace my bookkeeper or accountant?

No. It answers the questions between their reports, from live data. The bookkeeper still closes the books; the accountant still files. The agent removes the guessing in the thirty days between.

Can the agent move money?

Not the way we build it. Read access only, no posting, no transfers. That boundary is what makes it safe to connect to real accounts and what makes the security review short.

Do I have to change banks or accounting software?

No. Same bank, same books, same invoicing. The agent is a layer on top that can answer a question, connected through the APIs those systems already expose.

First published in The Humming Agent Brief, Shawn's twice-monthly newsletter for business owners. Subscribe there to get the next issue first.

Frequently Asked Questions

What is an AI CFO agent?

An agent with read-only access to your bank, payment processor and accounting system that answers cash questions from the live accounts: what is coming in, going out, owed to you and owed by you. It reconciles across the three systems rather than reading one report.

Does an AI CFO agent replace my bookkeeper or accountant?

No. It answers the questions between their reports, from live data. The bookkeeper still closes the books; the accountant still files. The agent removes the guessing in the thirty days between.

Can the agent move money?

Not the way we build it. Read access only, no posting, no transfers. That boundary is what makes it safe to connect to real accounts and what makes the security review short.

Do I have to change banks or accounting software?

No. Same bank, same books, same invoicing. The agent is a layer on top that can answer a question, connected through the APIs those systems already expose.

Ready to Transform Your Business?

See how AI automation can help scope practical workflow improvements and measurable business-case assumptions.

Book Operations Review
S

About Shawn Kercher

Shawn Kercher is the co-founder and CTO of HummingAgent, a Colorado AI automation agency that builds custom AI agents, voice receptionists, Private GPT deployments and workflow automation that clients own. He spent thirteen years as an engineer at Comcast Labs before founding the company, and writes The Humming Agent Brief, a twice-monthly newsletter for business owners.

Popular implementation paths

Turn this AI idea into a scoped workflow

Compare the related service pages and practical guides buyers most often use after reading our AI automation articles.