Seapoint

AI-native invoice pipeline that fuses email metadata (sender, inbox, full email body) with PDF extraction and bank transaction data — not just OCR of attachments but multi-source contextual parsing to reduce false matches and surface anomalies before approval.

Two‑way, near‑real‑time Xero sync combined with automatic transaction categorisation across every connected bank account — implies idempotent reconciliation logic, conflict resolution, and continuous mapping between bank feeds and ledger entries rather than periodic bulk imports.

Hybrid platform/partner architecture: Seapoint provides a single UX and product surface while outsourcing regulated rails and safeguarding to Modulr, Wealthkernel, and Yapily — they combine Open Banking AIS/PIS, e‑money issuance, and MMF investment products inside one orchestration layer.

Invoice capture from Gmail (connected mailboxes) with sender and full-body surfacing suggests use of persistent mailbox connectors + NLP/LLM pipelines that must handle noisy, multi-lingual email bodies, embedded images, inline receipts, and forward chains — a harder engineering problem than simple PDF OCR.

Product-level automation across payroll, treasury, cards, invoicing and bookkeeping in a single account implies sophisticated orchestration: event-driven workflows that trigger payments, update books, issue cards, and reconcile payroll with tax rules — all while maintaining auditability for accountants.