Platform overview
One AI Financial Data Graph for the whole office of the CFO.
FAQ
The questions finance teams ask first.
Straight answers on the ERP, approvals, accuracy, audit, and security. The same ones a controller or CFO asks in the first call.
- Does Sebastian replace my ERP?
- No. Your general ledger stays your system of record. Sebastian reads your ERP, CRM, HR system and warehouse into one AI Financial Data Graph™, does the accounting and planning work on top of it, and posts approved entries back to your ERP. No migration, no re-implementation.
- Who approves what the agents do?
- A person, every time. Agents prepare the work: matching, drafting, tie-outs, forecasts. Every entry is reviewed and approved before it posts. You see the proposal, the policy it cites and the sources behind it. A reviewer posts it to your ERP, and preparer and reviewer are kept separate by the system.
- How accurate is it?
- Accuracy is not a number we ask you to trust. Every action before the final one requires a human review: a person verifies the work before anything posts. Rule-bound work such as a three-way match or an ASC 606 schedule gives the same answer every time, and where judgment is involved the draft carries a confidence score so the reviewer knows how far to trust it. That makes Sebastian the most compliant natively built AI system for finance, with humans in the loop doing the verification.
- Which work is deterministic, and which takes judgment?
- A three-way match against the PO line, a subledger or bank tie-out, and an ASC 606 schedule off a signed contract are rule-bound: they run as rules and return the same figure on every run, because a controller would not accept anything less. Judgment is reserved for what genuinely needs reading, such as an ambiguous bank description or an accrual estimate, and those drafts carry a confidence score. The split matters at review time: on rule-bound work you are checking the inputs and the scope, and on judgment you are checking the reasoning.
- What happens when it gets something wrong?
- You reject a proposal rather than reverse an entry, because nothing posts without a person approving it. Every draft arrives with the calculation that produced it, the source records behind it and the policy it was decided under, which is what makes a wrong answer visible rather than merely plausible. Where judgment was involved, the confidence score says where to look first. The rejection is recorded against the draft, so a mistake that repeats is something you can see and correct rather than rediscover each month.
- Will it hold up in an audit?
- Yes. Every entry carries its calculation, its source records and the policy it was decided under. Source snapshots are versioned with control totals and a content fingerprint the server re-checks, and the audit binder assembles itself as the close happens rather than being reconstructed when the auditors ask.
- What does Sebastian connect to, and how is my data protected?
- Live integrations with QuickBooks, NetSuite, SAP, Workday, Salesforce, HubSpot, DocuSign, Ramp and Snowflake, and many more. Each customer’s data is isolated in its own workspace with field-level encryption, row-level access controls, segregation of duties and period lock.
- What accounting work do the agents actually do?
- Accounts payable with three-way matching down to the PO line, accounts receivable with an ASC 606 revenue schedule from each contract, bank and subledger reconciliations, accruals and prepaids drafted with the calculation and source records attached, and the close checklist itself: tasks open when the upstream work has tied out, and the audit binder assembles as the close happens.
- Does Sebastian do FP&A, or only accounting?
- Both, on the same graph. Forecasts, budget vs. actuals, flux analysis, and board reporting run on the same actuals the close produces, so the forecast updates as accounting lands instead of waiting for an export. Every figure in the deck drills to the transaction behind it.
- Can it follow our own accounting policies?
- Yes. Your revenue recognition treatment, accrual methodology, capitalization thresholds, materiality limits, approval chains, and chart of accounts are applied the same way every period, and each proposal names the specific policy it was decided under, so you are checking a decision rather than reverse-engineering it.
- How does it run the month-end close?
- As one workspace: tasks, owners, reviewers, dependencies, files, comments, and review activity in one place. A task opens when the work it depends on has actually tied out, not when a date arrives; owners are chased automatically; closed periods lock; and the evidence a reviewer relied on stays with the task.
- How is this different from an AI-native ERP or a close-management tool?
- Sebastian is neither a replacement ledger nor a checklist on top of one. It keeps your ERP as the system of record, holds the contract, invoice, PO line, payment, and journal entry behind every number on one AI Financial Data Graph™, and runs accounting and FP&A on it with agents that prepare the work and people who approve and post it.
Keep your ERP. Close and plan in one pane.
AI for accounting and FP&A on the ledger you already run. No migration. No rip and replace.
Agents at work · Matching an invoice to its PO line…