



Kinter accesses the financial data necessary to run your configured agents. This includes general ledger data, bank transactions, credit card transactions, payroll run data, and vendor records from your connected systems. You control which integrations are active and which agents have access to what data.
No. Kinter does not use actual customer data for model training. Your financial data is used only to process your agent workflows. This obligation extends to our model providers as well.
Kinter's infrastructure runs on airgapped AWS VPCs with data hosted in the United States. All data is encrypted at rest and in transit.
No. Any action that would modify your general ledger (i.e. posting a journal entry, categorizing a transaction, adjusting an accrual) requires explicit human approval through our structured review workflow. Agents prepare the work, your team approves it before anything is posted.
Kinter securely encrypts your credentials, which can be revoked at any time. ERP connections use OAuth 2.0 with scoped permissions. Bank connections use Plaid's tokenized infrastructure. Payroll connections use Finch's tokenized API.
Every agent execution produces a full audit trail: the source data it referenced, the logic it applied, the output it generated, and the approval or rejection by your team. This trail is immutable and exportable, designed to satisfy your external auditors during SOC 1 and financial statement audits.
Kinter follows least-privilege access principles. Access to customer data is restricted to a small number of authorized personnel for support and incident response purposes only. All access is logged and auditable.
Kinter conducts regular vulnerability scanning, code review, and will undergo annual third-party penetration testing and SOC audits as part of our compliance program.
If your security or compliance team needs more detail, reach out to security@kinter.ai. We're happy to walk through our controls, share our security documentation, or set up a call with our team.