Frequently asked questions.

BeanStack automates routine accounting work under policies you approve — document intake, extraction, bank reconciliation, journal posting, schedule maintenance, and the close checklist. Material, novel, and low-confidence items are held for review rather than posted silently. In Fully managed, a named BeanStack controller reviews those items; in Self managed, your controller does.

Yes — within approved policy boundaries, with the source evidence and reasoning attached to each one. Items outside those boundaries do not post silently; they escalate for named review. That is the difference from assistive tools: routine work completes instead of waiting in a suggestion inbox, and everything else stops and asks.

It replaces the repetitive, rules-based work — data entry, transaction matching, schedule maintenance, close checklist execution. Judgment calls like complex estimates, tax positions, and audit responses escalate to a named reviewer: your controller in Self managed, a BeanStack controller in Fully managed.

Start by connecting your bank, billing systems, documents, and current ledger. BeanStack can run over that ledger, or import your complete QuickBooks history onto ours in hours, not months. Then it reproduces one close in parallel against your current process in your first month — and you decide whether to keep your ledger, move, or stop.

No. BeanStack imports your chart of accounts and historical data from QuickBooks Online, and can run in shadow alongside your existing system for a full close before you cut over.

Your choice. Keep your ledger — BeanStack operates as a control plane over your existing QuickBooks or accounting system, applying the same policies, approvals, and evidence trail on top of the ledger you already run. Or move to ours — our migration AI moves your QuickBooks history — chart, transactions, and balances — onto BeanStack's ledger in hours, not months, with reconciliation and sign-off completed during the shadow month. Either way, a parallel shadow close shows what matched, what differed, and why — before you rely on the numbers.

All plans start with an AI-guided conversational onboarding — the AI assistant walks you through bank connections, chart of accounts, and posting rules step by step in the chat interface. For companies with more complex structures: a structured onboarding call where we map your specific workflows — revenue recognition patterns, entity structure, intercompany relationships, approval thresholds — before your first close.

Five categories: (1) Document extraction — reading invoices, contracts, statements, extracting the fields that matter. (2) Transaction classification — categorizing bank transactions, matching them to GL entries. (3) Journal entry generation — creating double-entry entries from extracted data and posting rules. (4) Revenue recognition — building and maintaining recognition schedules from contract terms. (5) Anomaly detection — monitoring for duplicate payments, unusual patterns, entries outside historical norms.

Extraction accuracy varies by document format and improves as the system learns your vendor formats. Every value includes a confidence score, and items below your configured threshold surface for review before posting — you confirm or correct a single field, not re-enter the document.

AI postings are reversible. You can review pending postings before they're finalized, correct a misclassified transaction, and the correction appears in the audit trail. The log shows the original AI decision and any human override. If you correct the same type of mistake repeatedly, the system updates its behavior for that pattern.

Yes. You define posting thresholds: transactions under a configured amount and confidence score post automatically; larger or lower-confidence items require approval. Separate thresholds by account type, entity, or transaction category. Intercompany transactions can require dual-approval by default.

Encrypted at rest (AES-256) and in transit (TLS 1.3), on cloud infrastructure in the US. Every organization's data is isolated at the database layer, and your data is never used to train AI models. See the security page for full detail.

No. Your financial data is processed by AI models for your benefit only — never used to train BeanStack's models and never shared with other customers. Our AI infrastructure providers process it under contractual data-processing terms that prohibit training on your data.

BeanStack supports multi-entity bookkeeping, multi-currency, and intercompany workflows. Full consolidated reporting is still being validated end to end. Tell us your entity structure and we will state exactly what is supported today.

Every posting, AI classification, and human override is logged with a timestamp, user identity, AI confidence score, source document, and the rule applied. The log is append-only and exportable at any time. Any number in your financials traces back to the originating document in one click.

Bank feeds via Plaid and AP via Bill.com are live today. QuickBooks Online is supported for chart-of-accounts and history import. Payments, payroll, and CRM connectors are rolling out, and BeanStack can read most systems that expose an API. A REST API is available for custom integrations — see the integrations page for current status.

For companies that have outgrown QuickBooks — multiple entities, complex rev rec, or high invoice volume — yes. Migration from QuickBooks Online takes hours — not a months-long implementation — and you can run one close in shadow before cutover.

For companies that need real ERP capability — multiple entities, complex revenue recognition, dimensional reporting — without a long implementation project: yes. BeanStack can run over your current ledger, or become the system of record after a reviewed shadow close. If your setup is unusual, we'll tell you honestly on a demo call.

One monthly fee for the whole finance department, scoped to your business — Fully managed if we run the department, Self managed if your controller runs the ledger. No seat minimums, no implementation fees, no per-transaction billing.

Better: a shadow close. Before you commit, BeanStack keeps your books in parallel with your current system and delivers one close you can score line by line against your own numbers. You evaluate the department on your own books, not a demo.

One monthly fee, scoped to your entity count, transaction volume, and operating model. Talk to us and we'll scope it.