The AI analyst with receipts

Your data, answered.
Provably.

Strata is the AI analyst for businesses that run on QuickBooks-class systems and decide on spreadsheets. Ask in plain English. Every number shows its work. A human approves everything it learns. Useful the day you connect it.

Book a 30-minute working session

Read-only by construction · Your own AI key · Your data never trains anyone's model

Strata · Live answer
Reference environment
Gross margin dropped in July — what happened?
Plan · edit it before it runs

Pull the monthly P&L by product line → compare July against June → decompose the margin change into price, cost, and mix → reconcile the drivers against the total.

Query 3 of 3 · read-only 638 ms · 24 rows
SELECT month, product_line,
       SUM(revenue) AS revenue,
       ROUND(100 * SUM(revenue - cogs) / SUM(revenue), 1) AS margin_pct
FROM   monthly_pnl
WHERE  month IN ('2026-06', '2026-07')
GROUP BY month, product_line
ORDER BY product_line, month;
Result · The margin bridge, June → July 3 drivers · fully reconciled
DriverJuly impactMargin points
Freight surcharges — hardware−$18,400−2.1 pts
New discounting — two accounts−$6,900−0.8 pts
Mix shift toward hardware−$1,700−0.2 pts
July change, total−$27,000−3.1 pts

July gross margin came in at — down from June — and it isn’t a pricing problem. Freight surcharges on hardware did most of the damage at ; new discounting on two accounts cost another −0.8, and the mix shift toward hardware explains the rest. The three drivers sum to the full −3.1 — nothing left unexplained.

Queriesqry_9b12d4 · qry_c803e5 · qry_a41b77 Sourcemonthly_pnl Datav218 · 6:02 AM
Reconciled — drivers sum to the change, zero unexplained “Margin” = revenue − COGS ÷ revenue · your definition, v3 As of 6:02 AM
the answer isn’t finished until you can correct it

Every answer arrives in this order — plan → query → number → proof → your correction.

The problem

No analyst. No trust. No time.

Your business generates plenty of data — it’s just trapped in the accounting system, the CRM, and nine spreadsheets, guarded by one overloaded “numbers person.” The figures below show their work.

25 hrs/wk

lost to manual data wrangling across ~10 disconnected tools.

Intuit QuickBooks survey · Aug 2024
40%

of CFOs don’t fully trust their own financial data.

BlackLine CFO survey · 2024
88–94%

of spreadsheets contain errors — and the month runs on them.

Panko · spreadsheet-error research
77%

of small businesses fear wrong numbers from AI — while most already use it.

Karbon State of AI in Accounting · 2025

The missing ingredient isn’t a smarter model. It’s trust — and trust isn’t a feeling, it’s machinery: receipts on every number, a human gate on every lesson, and a test suite that keeps right answers right.

What you get

Five advantages. One system.

Ask in plain English → trust every number → and it compounds, with your sign-off. Each piece exists somewhere. The system exists nowhere else.

01 / 05A real analystLive

Ask a question. Get a real answer — from your actual books.

Plain English in; your numbers out — revenue, margin, cash, aging — along with the exact tables it read to get there. No dashboard to build, no report-request queue, no warehouse project before the first question.

Elsewhere

AI commentary about a report you already ran, a dashboard builder and a lost weekend — or a real AI analyst locked behind a data warehouse you don’t have.

Asked at 9:14 AM · answered at 9:14 AM
What’s our real margin by product line?

Services carries the business at margin. Installation runs , and hardware is 11.2% after freight — freight wasn’t in your old spreadsheet’s formula.

02 / 05Shows its workLive

Every number arrives wearing its receipt.

Click any figure and see the exact query, the exact cell, and the exact version of the data it came from. Numbers you can put in front of your bank, your board, or your auditor — and defend when someone asks where they came from.

Elsewhere

A number you take on faith. Microsoft’s own documentation says Copilot “can produce incorrect answers” — and tells you to check.

Anatomy of a figure
$46,200
Queryqry_8f3c21 — the SQL, one click away CellSUM(open_balance) · rows 1–2 Dataversion 218 · synced 6:02 AM Accessread-only · your data identity
03 / 05Tested answersLive

Right answers stay right.

Answers you approve become your answer key — a test suite the product has to pass. Every change replays it, five trials per question, and a change that breaks a right answer doesn’t ship. And when the data can’t answer, Strata says so and names what’s missing — it never invents a number.

Elsewhere

Nothing stands between a model update — theirs, or their AI vendor’s — and your Monday report. An answer that was right yesterday can silently change.

Your answer key · 5 trials per question
Pass^5 · every change, every question Breaks 2 answers — doesn’t ship
04 / 05Learns your businessLive

Correct it once. A person approves. It’s learned.

“We exclude intercompany transfers from revenue.” Say it once — a reviewer you trust approves the lesson — and every future answer uses it. Your definitions, your fiscal calendar, your exceptions: an analyst that’s measurably better at your company every week.

Elsewhere

The chatbot on day 400 is the same generic bot as day one. Your corrections vanish when the chat window closes — and anything it does “learn,” nobody reviewed.

Curation queue · awaiting your reviewer
revenue includes intercompany transfers
revenue excludes intercompany transfers
Approve Fans out to → definitions · answer key · every future answer
05 / 05Your key, your dataLive

Your AI key. Read-only by construction. Nobody trains on your data.

You bring your own AI key, so your data runs under your agreement — and never trains anyone’s model. Access is read-only, enforced by the database itself, not by promises. Every AI call is metered on one ledger you can read.

Elsewhere

Their AI, their terms — training and retention buried in the terms of service, AI cost bundled where you can’t see it, and no say in which model reads your financials.

The posture, on one card
AI keysk-ant-••••••••••••-yoursYours
AccessSELECT only — database grantsRead-only
Trainingon your data, by anyoneNever
Spend$41.62 this month · per-call meteringOne ledger
How it compounds

Every correction makes the next answer better.

Four stages, running continuously — and a human gate in the middle of them. Nothing goes live automatically.

Capture

Every answer invites correction — a thumbs-down, a fix, a note. Nothing you teach it is lost.

Review

Corrections queue for a person you trust. Nothing goes live without their approval.

Promote

Approved lessons fan out — to your definitions, your answer key, and every future answer.

Prove

Every change replays the full answer key. Break one right answer, and it doesn’t ship.

0
Verified answers on file
0%
Answer-key pass rate
0×
Trials per question, per change
Live from our reference environment — a complete, fully synthetic customer simulation we run in production. No customer data appears on this page.

Everyone sells answers.
Ours get provably better every week.

Because the product isn’t the model — it’s the learning system around it.

Time to value

Useful the day you connect it.

Not a BI project. Not a quarter of consulting. A read-only connection and a first question, the same day.

Day 1

Connect and ask

Secure, read-only connect in minutes. First cited answers on your own numbers the same day.

Week 1

It speaks your language

Your definitions, your account names, your fiscal calendar — each one approved by you, then used everywhere.

Week 2

Your answer key exists

The answers you’ve marked correct become the test suite. The first Verified badges land.

Month 1

The routine runs itself

Scheduled reports arrive on time, watches flag what moved, and week one’s questions answer themselves.

The field

Each has a piece. Nobody has the system.

Reporting tools
Fathom · Syft · LiveFlow
Missing

AI that writes commentary about a report you already ran. No questions, no receipts, no learning.

BI + copilots
Zoho Analytics · Power BI Copilot
Missing

Microsoft’s own docs say Copilot “can produce incorrect answers.” And the data model behind it is your unpaid job.

AI analysts
ThoughtSpot · Zenlytic · Cortex
Missing

Real chat, real lineage — behind a data warehouse a small business doesn’t have and shouldn’t need.

The ledger’s own AI
Intuit · Xero
Missing

Locked to one ledger, upgrade-gated — and grading its own homework.

Only Strata has the full loop
No warehouse required Every number cited Tested against your answer key Human-approved learning Your own AI key

And the system is what compounds — every answer, correction, and approval makes the next one better.

Who it’s for

Built for the people who answer for the numbers.

10–500-employee businesses on QuickBooks, Xero, or Zoho — multi-entity, board- or lender-accountable, spreadsheet-weary — and the advisors who serve them.

Owner

“Know your numbers without becoming the numbers person.”

A cited answer on your own books, in the first session.

CFO · Controller

“Every number defensible — to the board, the bank, the auditor.”

Click a figure: the query, the data snapshot, the definition behind it.

Fractional CFO · Firms

“Serve twice the clients with the judgment you already have.”

The two-day board pack, as a scheduled report with receipts.

Ops leader

“Stop reconciling ten tools. Ask one question.”

One governed layer over the ledger, the CRM, and the spreadsheets.

The posture

Safe to connect everything.

Read-only by construction

Enforced by database permissions, not promises. It can look at your books; it can never write to them.

Your own AI key

Bring your own key, keep your own terms. Your data never trains anyone’s model — ours included.

Every AI call metered

Spend sits on one ledger you can read. No silent model swaps — a downgrade is always labeled.

A human approves every lesson

Everything it learns passes through a review queue owned by someone you trust. Nothing goes live automatically.

A full audit trail

Every query, decision, and approval is on the record — who, what, when, and on which version of the data.

Honest refusals

When the data can’t answer, it says so and names exactly what’s missing. It never invents a number to be helpful.

The next step

See it on your own books.

A 30-minute working session: connect read-only — or start from our reference environment — ask your hardest question, and watch the answer show its work.

01

Tell us what you run on

QuickBooks, Xero, Zoho, a warehouse, or a pile of spreadsheets — all fine.

02

A 30-minute working session

On your data, read-only — or on the reference environment if you’d rather look before connecting.

03

Keep it if it earns it

Nothing installed, nothing written, nothing locked in. If it doesn’t earn its keep in week one, walk away clean.

Read-only connect · No obligation · Your own AI key from day one

Request a working session

No list, no drip campaign — a person reads this and replies.

Request received.

A person will read it and reply — with a proposed time, not a drip sequence.