The board pack. The signed number. Scroll, and step through to what stands behind it.
Reconciled
Traced to source
Recomputed
Materiality
Coaching
Provenance
Pushback that sharpens
Challenged
Improve the close
Audit-ready
Covenant headroom
Signed off
Defensible
Huxley
Close Intelligence
Every figure recomputed from source. Deterministic, not guessed.
Traced to its origin, with materiality and coverage on the drivers
A reviewer that challenges the close before you sign
Know your numbers hold
Huxley
Close Intelligence
Know your numbers hold.
Enter
Scroll
The month-end close where every number traces to its source. New to the seat or twenty years in, the answers are ready before you walk in: variances explained, covenant and control gaps flagged early, a record clean enough for the owner, the lender, and the auditor.
If you can export it, Huxley reads it. Any system, any spreadsheet, a plain CSV.
The expectation has changed
The finance engineer.
Not a new title. A new expectation inside finance. Reconciliation that runs before the close. Reporting validated before the board pack. Every exception owned, every step on the record. Less heroics at ten at night, more workflows the business can rely on. That shift is the whole reason Huxley exists.
Checks before the close
Automated reconciliations, variance thresholds, and exceptions caught while there is still time to fix them.
✓ Cleared
Validation before the board pack
Tie-outs, policy and disclosure checks, and one source the pack reads from, so it cannot contradict itself.
✓ Validated
Every exception owned
Ranked by impact, each with a named owner and an approval path, and its resolution kept on the record.
✓ Resolved
Controls on the record
Role-based access, segregation of duties, and a full audit trail built as the work happens, not reconstructed later.
✓ Captured
Board pack ready. Trusted, complete, defensible.
Build a close the business can rely on, not one it has to take on faith.
The report that drives decisions
A report that describes the past won't drive a decision.
Most close packs explain what happened, then wait for questions. Huxley is built to answer the next one, while the decision is still live.
Describes the past · most reports
✗Reports what happened, then waits for questions.
✗Built from the chart of accounts, not the agenda.
✗A variance with no owner, no cause, no next step.
✗Forty lines reported. Nothing communicated.
✗Lands days after the decisions were already made.
✗"Performance was mixed" is the closest thing to an action.
Drives decisions · Huxley
✓Leads with the verdict, and the one thing to do about it.
✓Built around the decisions leadership actually owns.
✓Every variance carries its driver, its impact, and the ask.
✓Three headlines up front; the detail one click away.
✓Ready while the decision is live, and it traces to the close.
✓Every section ends with one ask: approve, redirect, escalate, or monitor.
A feel for the close, in a few screens.
The readiness posture, a board-grade verdict with its sources, covenant headroom, controlled sign-off, and the relationships you inherit. Not a tour, just the shape of it.
Tap for sound
One close. Two seats. Step into either.
A financially fluent CFO and a non-financial owner need the same month to read completely differently. Huxley keeps one set of reconciled numbers and renders the seat each person needs. Try it below, this is the actual idea, live.
May 2026 close · demo data
Same reconciled numbers underneath. Flip the seat, not the truth.
CFO view
Density for someone who reads columns
Everything, in the order you check it.
Close-health at a glance, the checklist line by line, what is reconciled and what is blocked, covenant headroom, the figures behind every number.
+ Close-health ring and KPI variance against budget
+ Checklist with owners, status and due dates
+ Flags, covenants and the full statements underneath
Owner view
Translation for someone who acts on trust
The plain truth, in one sentence.
A verdict you can read in five seconds, the one or two things that actually need you, and a clear line back to the books when you want it.
+ "Your business made money, and the books are clean"
+ What needs you, in plain English, never jargon
+ Every number traces back to a reconciled account
same close, told two ways
It reviews your close the way the auditor will, before they do.
At every stage, the journal entries, the variance commentary, revenue recognition, the cash forecast, the covenants, and the sign-off itself, Huxley raises the questions a careful reviewer asks first. Each is ranked by what is material, carries the figure behind it, and ends with the fix. Nothing reaches the board that you have not already been challenged on.
One of ten reviewer reads across the close. Here, the 13-week cash forecast: the floor it breaches, the single collection it leans on, and the stress test it never ran.
A board pack assembled from many files can contradict itself.
The summary is built Monday from one export, the appendix Wednesday from another, and the data refreshes in between. Now the same figure reads one number on the summary and another in the appendix. No one made a mistake, which is what makes it dangerous: the moment a director catches one number that does not tie, every number is suspect. Huxley shuts that door by construction. Every section reads from one close, so the pack cannot disagree with itself, and the judgment on the numbers stays yours.
The black box
"Trust us, the number is right."
× The work was done, but the reasoning lives in someone's head, not on the page.
× Every variance becomes another round of questions the auditor has to chase.
× When a control gap slips through, nobody sees it until year end.
× Trust rests on the reputation of whoever is presenting, not the evidence.
The glass box
"Here is the number, and here is why."
✓ Every material movement carries a written, reviewed explanation.
✓ The evidence trail is built as the work happens, not reconstructed later.
✓ Control gaps surface during the close, while they can still be fixed quietly.
✓ Trust rests on what anyone can inspect, so it survives a CFO change or an audit.
The deliverable
The board pack, built from the close it reports on.
Not a deck rebuilt by hand each month. Huxley assembles the pack from the same reconciled close: the recommendation up front, the evidence behind it, and every figure one click from its source. It reads the same at month, quarter or year, and prints clean.
This part is live. Switch between the month, the quarter and the year to date, or click any number to trace where it comes from. The same reconciled close underneath, whichever way you read it.
Structured like a decision
Situation, insights, recommendation, evidence, the ask. The reader gets the point first, then the proof, not a wall of tables.
Charts that tie out
A trailing-year trend, an EBITDA bridge from plan to actual, and a net-income walk by line. Each reconciles to the dollar and traces to the ledger.
Month, quarter or year
One toggle re-reads the whole pack on the basis a director asks for, without rebuilding a thing. Save it as a clean PDF in a click.
Answered from the numbers
The questions a board actually asks.
Not "what happened," which the statements already show. The harder ones, each answered from the close itself, with the working underneath and nothing invented.
Are we creating value?
Return on capital against the cost of that capital, by entity and for the group. The spread that says whether growth builds value or spends it down.
How good is our plan?
Forecast accuracy by driver over the trailing year. The lines that keep missing in the same direction, which is a planning gap, not noise.
What is the downside?
Named, versioned scenarios. A volume drop or a cost rise, run through to the covenant, reproducible from its inputs every time.
Where does cash come from?
Cash generated and consumed by unit, and what is tied up in receivables and stock, so the group total has faces behind it.
Are our initiatives paying off?
Each project's committed benefit against what the ledger actually shows. A reported saving the numbers do not back is flagged, never counted.
Will we hold our covenants?
Headroom on every covenant with the forward test, so a breach is seen while there is still time to act on it.
The whole close, in one defensible workspace.
Close and tie out
Import the trial balance from any ERP. Three statements tie, intercompany eliminates, the calendar tracks every task and sign-off through to day five.
Document and explain
Variance commentary, segregation-of-duties controls, and a full evidence trail, written as the work happens. Board pack and audit binder export in one click.
Prove progress
A visible record of how the function improved over time, including the months it slipped. The artifact that travels up to the CFO and the sponsor.
Is this AI, and can you trust it?
AI drafts the words. It never decides the numbers.
The fair question every finance leader asks. The honest answer is a governance boundary. On one side, AI gathers context and proposes language. On the other, a person reviews and approves, and every decision is logged so it can be reconstructed for an audit. The judgment stays human, and the trail behind it stays complete.
Step 1System
Surface
Huxley flags the material movements, control conflicts, and variances that need an explanation.
Step 2AI assist
Draft
AI turns rough notes into clear, board-ready language. It never invents a figure; it only explains the ones already there.
Step 3Human judgment
Review and approve
A named person reviews the draft and approves it. Nothing reaches the board pack without a human signing off.
Step 4Permanent trail
Log for audit
Every approval, edit, and commentary is timestamped and attributed, so the decision can be reconstructed months later.
⊣
Low autonomy, by design
The bigger the decision, the less you want it automated, and a board-ready close is the highest-stakes work a finance team ships all month. So the AI gets room to surface and draft, and none to decide: it proposes, a named person approves, and every figure traces to its source. That boundary is what makes the result safe to put in front of a board.
Your systems record the numbers. Huxley makes them trusted.
There is a chain that runs through every finance function. Huxley lives in the gap the tools you already own were never built to fill. Most finance AI enters that chain at the top, at the forecast, and trusts the close beneath it; Huxley starts at the close, where the trust is actually built.
Your ERP
Records what happened
The system of record. It holds the transactions and produces the raw numbers, but tells no story and proves nothing on its own.
Planning tools
Look forward
Budgets, forecasts, scenarios. Built to look ahead, not to close, document, and explain what already happened.
Huxley
Closes it, proves it, communicates it
Turning a finished month into documented, defensible, board-ready communication. Enterprise suites serve this, but are built for large teams and long rollouts.
Board, auditors, sponsor
Have to trust it
The people who depend on the numbers and hold finance accountable. Trust is built or lost right here.
The sweet spot is the team that has outgrown spreadsheets and email but is too lean for heavy enterprise close software. That is exactly the band that carries the scrutiny of a big company, an audit, a board, often a PE sponsor, without the infrastructure of one. The expectations are enterprise. The tooling is a spreadsheet. That mismatch is where a small undetected error becomes a year-end surprise. Huxley closes that gap.
I spent ten years in the auditor's chair, and I was handed a spreadsheet with no story behind it.
I would walk into companies everyone believed were well run, and at audit time all I had was that spreadsheet. No explanation, nothing I could use. So I went back to the client, again and again, looking for the reasoning behind the numbers.
Then I sat on the other side of the table. A Big Four audit foundation, then more than fifteen years as a CFO across public-company, PE-backed, and family-owned businesses. I did not just ask for better reporting, I built it, architecting the BI and data platforms myself, to the point where external auditors trusted my model over the ERP.
The lesson from both chairs was the same. I had sat in meetings where a controller closed the month cleanly, every figure right, and still lost the room to blank stares. The work gets done, it just never gets communicated, and that costs finance the trust it has earned. Huxley is that knowledge built into a tool. I named it after my son. I built it because I have watched too many good finance people miss bedtime while the close ran at ten at night. Better documentation is about trust, and it is about finance teams getting their evenings back.
Danny Straten
Founder · former Group CFO
Photography by the founder
Straight answers to the obvious questions.
Does this replace my CFO, controller, or auditor?
No, it equips them. The work and the sign-offs stay with your team. Huxley makes what they do documented and defensible, so the numbers hold up when the owner, the lender, or the auditor asks.
Is the AI deciding the numbers?
No. AI drafts the wording and flags what needs a second look; a named person reviews and approves every figure before it moves. The judgment stays human, and the record proves who decided what.
Whose data is it, and where does it live?
Yours. Every number traces back to your own source, and you can export the full record whenever you want. Huxley documents your close; it is not a system of record that owns it.
Will my data be fed to an AI, and is it kept private?
AI only drafts the wording for numbers that already exist; it never decides a figure, and your data is not used to train any model. The numbers stay yours, they trace to your own source, and you can export the full record at any time.
See where your close stands today.
A short, honest read on how documented and defensible your month-end is right now, and the few things that would move it most.
