Why suddenly everyone is rediscovering SQLite
The SQLite renaissance is the cluster of trends that, between roughly 2022 and 2026, made the in-process database fashionable again after twenty years of Postgres-as-default. The main drivers:
- Edge compute — Cloudflare Workers, Fly Machines, Vercel Functions. The compute moved closer to users, and the database needed to follow. SQLite ships inside the runtime.
- Litestream / LiteFS (Ben Johnson, Fly.io) — added streaming replication and synchronous replication on top of plain SQLite, removing the single-server objection for many workloads.
- Turso — distributed SQLite-as-a-service with global replicas, libSQL fork with vector search.
- Cloudflare D1 — managed SQLite per Worker.
- Local-first software (Linear, Ink & Switch research, etc.) — apps that work offline and sync optimistically. The local database must be embedded.
- RDS/Aurora cost fatigue — managed Postgres bills shocked enough teams that 'just put it on the box' started looking like an upgrade.
- Modern SQLite features — STRICT tables, JSONB, vector search experiments, FTS5 trigram tokenizer, generated columns, native UPSERT.
Self-reference: Pippa is a local-first product by design. The data layer lives on Dad's machine, not on a SaaS. SQLite + JSONL is the embodiment of that choice — the same renaissance pattern, applied to one human's AI daughter instead of a startup.
This quest will teach you SQLite as a serious data layer — not a toy, not just a test fixture, not just a mobile cache. The renaissance is real because the use cases are real.