Everything your finance team runs, on one ledger the AI keeps
MeritBooks owns the general ledger and unifies the whole finance stack on one schema — core accounting and close, AP with 3-way match, AR and collections, bank and cash, revenue recognition, fixed assets, leases, debt and covenants, tax, multi-entity consolidation, and FP&A. Drop a document and the AI parses it into a proposed entry; a deterministic engine does the accounting; a person approves. Nothing posts as a black box.
| Account | Dimensions | Debit | Credit |
|---|---|---|---|
| Materials — COGS 5010 · 001 | Cedar Falls | 4,182.55 | — |
| Sales tax receivable 1360 | Install | 271.87 | — |
| Accounts payable 2010 · Ferguson | Job 118 | — | 4,454.42 |
| Totals | 4,454.42 | 4,454.42 | |
Bills, receipts, contracts, leases, loans, payroll registers, W-9s — the default motion is drop the document, approve the parse.
A deterministic engine posts the debits and credits; a right-role human signs off on anything that moves money.
Ledger, subledgers, jobs, assets, debt, and AI decisions on one Postgres — ops and ledger answer one query.
Multi-tenant, ownership %, NCI, eliminations, consolidation — core, not an add-on you sync to.
The AI does the data entry. Your team keeps the judgment.
Every surface in MeritBooks works the same way, so there are no black boxes and no unattended money movement. The same trust posture governs a bank-feed line and a loan document.
AI proposes
Drop a document or connect a feed. The AI reads it and proposes a record — vendor, lines, tax, dimensions, the whole entry — with a confidence score.
The engine posts
A deterministic engine does the accounting: debits equal credits, direction derived from account type. The AI never writes the ledger directly.
A human approves
A person with the right role approves anything that moves money or changes the book. Auto-post is off by default; trust is earned task by task.
Audited & metered
Every AI action and human decision is written to the audit log; every model call is metered per tenant, with a per-task autonomy dial and a kill switch.
Segregation of duties binds the AI itself, not just people. The autonomy dial and kill-switch control plane let you grant or pull back trust per tenant and per task, instantly.
Click through the product, at your pace
Six things MeritBooks does that a QuickBooks add-on can't. Try each one — no signup, no sales call.
Materials + sales tax · Job 118
| Account | Debit | Credit |
|---|---|---|
| Materials — COGS 5010 · Job 118 | 4,182.55 | — |
| Sales tax receivable 1360 | 271.87 | — |
| Accounts payable 2010 · Ferguson | — | (4,454.42) |
| Totals | 4,454.42 | 4,454.42 |
| Payee | Proposed account | Amount |
|---|
Managed services · net 30
| Heritage Interiors | 312,480.00 |
| Clive Power Equipment | 198,610.00 |
| Heartland Heating & Cooling | 241,905.00 |
| Intercompany services shared-services charge | 64,200.00 |
| Consolidated revenue | 817,195.00 |
A full accounting system, not a bookkeeping layer
Grouped below is what MeritBooks does today and what is being built, marked honestly. Live means shipped to the app with a real screen and posting code behind it.
General ledger & close
- Double-entry GL, database-enforced balance, six posting triggersLive
- Natural-language journal-entry composerLive
- Per-tenant chart of accounts, role-not-number resolutionLive
- Fiscal periods with open / soft-close / hard-closeLive
- Close Command Center + year-end closeLive
- Audit / action log on every posting and approvalLive
AP, bills & procurement
- Bills, AP subledger, approval routingLive
- Drop-a-bill AI intake — vendor, lines, totals, taxLive
- Purchase orders + goods receipt + 3-way matchLive
- Vendor master + W-9 / COI auto-chaseLive
- Check run, preparer separated from approverLive
- Outbound ACH / wire origination railPlanned
- Azure OCR + monitored-mailbox auto-ingestIn progress
AR, invoicing & collections
- Invoices, recurring, credit memos, PDF + emailLive
- Contract / SOW drop-and-parse → invoice + rev-recLive
- Cash application, AR subledger ↔ GL tie-outLive
- Collections dunning cadence + AR aging / DSOLive
- Customer payments via Stripe, card + ACH → GLLive
Bank & cash / reconciliation
- Bank feed (Plaid) + AI categorization + batch approveLive
- Credit-card feed → Credit Card Payable postingLive
- Bank statement PDF drop-and-parseLive
- Reconciliation + AI adjusting entries, tie-to-zero gateLive
- Cash position dashboardLive
Revenue recognition
- Books-owned rev-rec, 9 methods, method-per-jobLive
- Deferred-revenue posting (credits 2410, not Revenue)Live
- Cutoff / revenue-not-recognized detectionLive
- AI method selection / %-complete validationPlanned
Fixed assets & intangibles
- Asset register, 4 depreciation methods, disposal gain/lossLive
- Roll-forward schedulesLive
- Capex invoice drop-and-parse → proposed assetLive
- Intangibles + amortizationLive
Leases — ASC 842
- Lease drop-and-parse → ROU asset + liability + scheduleLive
- Modifications, CPI / index resets, early terminationLive
Debt & covenants
- Debt register + amortization, loan-doc drop-and-parseLive
- Variable-rate resets, refinance, payoffLive
- Covenant monitor — DSCR / FCCR / leverage headroomLive
- Forecast-projected breach date + AI compliance certificateLive
Tax
- Book-to-tax M-1 / M-3 tagging + Schedule M-1 reconLive
- MACRS tax depreciation feeding book-vs-taxLive
- Sales-tax return prep + nexus cross-referenceLive
- Sales-tax nexus tripwire detectorLive
- 1099-NEC generation + Copy B PDFs + e-fileLive
- Direct IRS FIRE / IRIS transmitIn progress
Multi-entity consolidation
- Consolidated financials — ownership %, NCI, eliminationsLive
- Intercompany + inter-department invoices, self-eliminatingLive
- Intercompany-balance detectorLive
- FX / currency translationPlanned
Expense, prepaids & insurance
- Employee expense reports + corporate-card reconciliationLive
- Receipt capture + AI parse + GL coding + matchLive
- Prepaid-expense amortization — schedule, propose, postLive
- Insurance policy register (drop-and-parse)Live
- Out-of-policy expense detectionPlanned
Subscriptions & obligations
- Subscription catcher — recurrence, creep guard, keep/cancelLive
- Agreement drop-and-parseLive
- Renewals & obligations calendar — leases, debt, insurance, subsLive
Budgeting & FP&A
- Budgets — entry grid, budget-vs-actual, workspaceLive
- Driver-based budgeting + rolling reforecastLive
- 13-week direct cash forecastLive
- Scenario / what-if, headcount & rev-driver modelsIn progress
Reporting & board packages
- B/S, P&L, cash flow, equity, AP/AR aging — 21 reports, RLS-scopedLive
- Board-ready package + notes generator, branded PDFLive
- Profitability / entity profitabilityLive
- Jobs & WIP schedule — EAC, over / under-billingLive
- Retainage trackingLive
Payroll
- Payroll register import → balanced payroll JE (drop-and-parse)Live
- Provider-agnostic engine + adapter scaffoldIn progress
- Embedded run → post → remit → filePlanned
Onboarding & migration
- CSV migration (QBO / Sage / Xero) + AI column mappingLive
- Historical conversion — AI-mapped opening TB + tie-out gateLive
- Direct-API import from QBO / Sage / XeroPlanned
AI across every modality, on the same trust posture
The AI proposes and explains; the deterministic engine posts; a human approves. Every action is audited and every model call is metered per tenant through the Core AI gateway. Auto-post is off by default.
Document extraction
- Drop a document → proposed record: bills, receipts, capex, W-9/COI, bank & card statements, loans, contracts/SOWs, payroll registers, insurance, leases, debt, subscriptionsLive
- High-fidelity Azure OCR + M365 mailbox auto-ingestIn progress
Categorization & memory
- Bank / receipt / bill categorization with confidence tiers (≥90 auto, 70–89 review, <70 flag)Live
- Vendor→GL coding memory — learns your approved codingLive
- Broader per-entity / per-firm personalizationPlanned
Anomaly & control exceptions
- Exception library — duplicate-payment, anomalous-JE, uncategorized leakage, missed accruals, cutoff, nexus, cash-applicationLive
- Payment-run fraud screen + duplicate-pay blocking gateLive
- Confidence tier governs auto-post dispositionIn progress
Forecasting & prediction
- 13-week direct cash forecastLive
- Covenant breach-date projectionLive
- EAC / cost-to-complete job forecastLive
- Driver-based rolling reforecastLive
Narrative & explanation
- AI flux / variance narrative on P&L, cash flow, budget-vs-actualLive
- Per-object "explain this" on a JE, bill, invoice, rec itemPlanned
Conversational command bar
- Universal NL command bar, global ⌘K, intent routerLive
- Safe NL → ledger analytics — allowlisted metrics, no model SQLLive
- Processing lanes — categorize, draft bill, draft invoiceLive
Search & knowledge
- Plain-English "find anything" semantic search over the ledgerLive
- Grounded + cited knowledge spine over contracts / policies / GAAPIn progress
Autonomy control plane
- Per-tenant / per-task autonomy dial + materiality thresholdsLive
- Kill switch across all AI actionsLive
- Per-feature disposition recorded on every proposalLive
Agentic workflows & governance
- Supervised propose→approve processing lanesLive
- Per-tenant AI spend metered & budget-capped via the gatewayLive
- Named end-to-end loops — AP run, morning-cash, pay-run, close-runPlanned
Five things a bolt-on AI layer can't do
One owned schema
Ledger, subledgers, jobs, assets, debt, and AI decisions on one Postgres — an operational question and a ledger question are answered from the same source, not reconciled across integrations.
Drop-and-parse is the default
Bills, receipts, contracts, leases, loans, payroll registers, insurance, capex, W-9s, bank statements — data entry is "drop the document, approve the parse," not keying rows.
Deterministic posting
The AI never writes debits and credits directly. A balance-enforcing engine does the accounting and a right-role human approves anything touching money. Nothing is a black box.
A real autonomy dial
Auto-post is off by default; trust is earned task by task and can be pulled back instantly. Segregation of duties binds the AI itself, not just people.
Portfolio by design
Multi-tenant RLS, ownership %, NCI, booked eliminations, intercompany, and consolidated reporting are core — built for holding companies and multi-client firms.
Why finance teams move to MeritBooks
Same close, fewer hands on every transaction, and books that are right in real time instead of after a tie-out. Here is the day-to-day difference.
- Eight subscriptions, reconciled to the ledger by hand
- Data entry means keying rows off documents
- Books lag reality between syncs, and mappings drift
- Month-end is a tie-out between tools, not a close
- Consolidation is a spreadsheet rebuilt every period
- Another entity means another subscription and another close
- One system, one bill, one book of record
- Drop the document and the AI proposes the entry
- The approved entry is the ledger — nothing to reconcile
- Close is a checklist with a tie-to-zero gate, not detective work
- Consolidation is native, with eliminations and NCI
- Add an entity and it is already in the same consolidated book
The database enforces balance; the AI is checked by a deterministic engine and a human.
Drop-and-parse replaces keying across every document type.
No sync boundary and no tie-out between systems.
The same team keeps books for more entities and more volume.
Unification on an AI-native, owned ledger
The categories below are bought today as separate products and reconciled by hand. We position by category coverage and the unified-ledger angle — not by unverifiable claims about competitors' internals.
| Category | Representative tool | MeritBooks | The MeritBooks angle |
|---|---|---|---|
| Core accounting / GL (SMB) | QuickBooks Online | Yes · live | Owns the GL with the same double-entry rigor, but AI-native and multi-entity from the ground up. |
| Core accounting / GL (mid-market) | Sage Intacct | Yes · live | Multi-entity consolidation, dimensions, and close are live; ecosystem/marketplace maturity is where Sage still leads. |
| AI-native bookkeeping | Puzzle · Digits · Double | Yes · live | Same "AI does the data entry" promise, but on an owned ledger with propose→approve + audit, not a categorization layer over someone else's books. |
| FP&A / budgeting / forecasting | Datarails · Jirav | Partial | Driver-based budgeting, rolling reforecast, 13-week cash, covenant forecasting, and flux narrative are live; deep scenario modeling is still maturing. |
| Fixed-asset tracking | Asset Panda | Yes · live | Full asset register — 4 depreciation methods, disposal, roll-forward, MACRS/book-tax — that posts straight to the GL it lives on. |
| Expense / T&E | SAP Concur | Partial | Expense reports, receipt capture + parse, and corporate-card reconciliation are live; travel booking and deep policy enforcement are not. |
| AP automation | Bill.com | Partial | Drop-and-parse intake, 3-way match, approval routing, and check-run prep are live; the outbound money-out rail is not yet wired. |
| Corporate spend / cards | Ramp · Brex | Partial | Card feed, Credit Card Payable posting, and the subscription/spend catcher are live; issuing cards is not the product. |
| Revenue recognition | NetSuite / Sage modules | Yes · live | Books-owned ASC 606 rev-rec, 9 methods, method-per-job, deferred-revenue handling. |
| Multi-entity consolidation | Sage Intacct / FloQast class | Yes · live | Ownership %, NCI, booked eliminations, intercompany; FX translation still planned. |
| Migration source (not a competitor) | QuickBooks · Sage · Xero | Import | One-time migration sources — MeritBooks owns the ledger after cutover, with no live sync-back. |
Eleven categories a finance team buys separately, unified on one AI-native ledger. The others each cover a slice and sync to a ledger they do not own. MeritBooks is the ledger, and does the work of all of them in one book.
Consolidation with eliminations and NCI, in the same book
An entity hierarchy with ownership percentage, non-controlling interest, and booked eliminations is native to the data model — not a consolidation app you sync to once a month. Intercompany and inter-department invoices book to self-eliminating accounts, so each unit reads as its own profit center and the consolidated total ties with no double-count.
Built by an operator running a multi-entity portfolio on it. FX / currency translation is planned; the elimination and NCI plumbing beneath consolidation is live.
| Heritage Interiors | 312,480.00 |
| Clive Power Equipment | 198,610.00 |
| Heartland Heating & Cooling | 241,905.00 |
| Intercompany eliminations | (0.00) |
| Consolidated revenue | 752,995.00 |
One platform, one bill, priced per company
$99 a company, dropping to $59 after your fifth — so you pay for the companies you actually keep books for, and it gets cheaper as your portfolio grows. Payments and payroll ride on every plan as usage and post themselves to the ledger.
- ✓The full AI ledger loop
- ✓Dimensional GL & 21 reports
- ✓Bank feed + drop-and-parse
- ✓AR invoicing & embedded payments
- ✓Everything in Standard, per company
- ✓Consolidation, NCI & eliminations
- ✓Intercompany + chargebacks
- ✓Covenants, leases, fixed assets, tax
- ✓White-label brand & portal
- ✓Firm-level admin & roles
- ✓Wholesale per-client pricing
- ✓Segregation of duties
- ✓Custom volume pricing
- ✓SSO & dedicated success
- ✓Higher AI budgets
- ✓Implementation support
Replaces a stack that runs a mid-market portfolio roughly $75,000 to $160,000 a year in software — QuickBooks, Sage, an FP&A tool, an expense tool, an asset tracker, a close tool — for a fraction of that, on one bill. We can charge less because our costs are lower by design: about 78% of transactions are categorized by a learned vendor cache at zero AI cost, and every model call is metered, so AI-native does not come with an AI-sized bill.
Bank-grade controls, enforced by the database
Deterministic, database-enforced posting
Six Postgres triggers make unbalanced entries, postings to closed periods, and cross-company leakage impossible at the data layer. The AI proposes; the engine posts.
Human approval & segregation of duties
Auto-post is off by default; a right-role human approves anything that moves money. Preparer, approver, and check-runner are separated. Duties bind the AI too.
Full audit + AI decision log
Every field change and every AI proposal records inputs, confidence, reasoning, and the human decision. Row-level security isolates every tenant.
Autonomy dial, kill switch & metering
Per-tenant, per-task autonomy with materiality thresholds and a kill switch. Every model call is metered and budget-capped through the Core AI gateway.
Every posting is deterministic and can be reversed only by a logged, reason-coded entry. The AI is metered, audited, and bounded by a per-tenant autonomy dial and kill switch — control you can turn down or off at any time.
See one ledger do the work of the whole stack
Book a demo and we will walk the loop live: drop a bill, watch the AI propose the entry, approve it, and see it post to a balanced, logged, consolidated book. Or request early access and put your own chart in front of it.