Our standard stack is deliberately conservative: we choose technologies we know can support a case management system for the long haul, because the retention obligation for some files runs for decades.
Backend in Postgres with RBAC. Postgres is our standard database for case management systems: open source, widely supported, with row-level security for multi-tenant separation. The role and permissions model lives in the database itself, not just in application code, so even an ad-hoc query goes through the same access control as an end user.
Frontend in Astro or Next.js. For the case workspace and the client portal, we choose Astro or Next.js, depending on the interactivity and SEO requirements of the portal. For heavy data-table screens, we use React with TanStack Table; for the client-facing side, we deliberately keep things simpler.
Document store on S3-compatible storage with KMS. Documents are held in object storage (AWS S3, Google Cloud Storage, Azure Blob or an EU-hosted alternative). Encryption at rest uses KMS keys that you manage. For on-premises, we use MinIO. Versions are stored immutably with object lock where the retention obligation requires it.
Search, OCR and speech-to-text. For search, Postgres' tsvector for smaller environments, OpenSearch or Meilisearch for larger ones. Documents are OCR'd using a combination of Tesseract and cloud OCR. Audio (statements, phone recordings) is processed by Whisper or a comparable model, on-premises too if the privacy context requires it.
Authentication via SSO and MFA. Single sign-on on your existing identity provider (Microsoft Entra, Google Workspace, Okta, KeyCloak), with MFA mandatory for all staff. For clients in portals, magic links or DigiD; for municipal chain partners, eHerkenning.