One database built from scratch — roughly 283,000 lines of code — that does the work of six specialized systems. Business tables, flexible documents, instant lookups, relationship maps, AI search, and live event feeds all live in one engine, saved together in all-or-nothing updates, and reachable by the software you already run, because it speaks the same languages those databases speak.
Underneath everything sits a single storage engine with one safety journal and one rulebook for saving data — proven to survive crashes without losing a saved record. There is no glue layer and no overnight sync job: a document, a table row, and an AI-search entry sit side by side in the same store.
Standard business tables and reports, in the same language PostgreSQL and MySQL speak — so tools and applications built for those databases, including some of the biggest open-source products in the world, run without changes.
Flexible records that don't fit neat rows, stored MongoDB-style — tested question by question against the real MongoDB to confirm the answers match exactly.
The instant-answer layer. It speaks the languages of Redis and Memcached — the standard tools for caching and background job queues — and real job-queue software runs on it end to end.
Maps of how things connect — customers to orders, people to companies. Ask relationship questions the way Neo4j users do, and the same data is still readable as ordinary tables, because it's one store.
The memory behind AI features: "find me things similar to this" search, answered in milliseconds — the layer the platform's built-in AI uses to remember and recall.
A built-in stand-in for Kafka, the industry system for live event feeds — send events with standard tools, then search them like any table the instant they arrive.
Compatibility claims are cheap, so SynergyDB doesn't make them — it measures them. The same set of operations is run against SynergyDB and the genuine original database side by side, and the answers are compared down to the last byte.
The discipline behind the numbers: every headline claim can be re-run from tests kept alongside the code, and anything that can't be re-verified gets cut. That verify-first posture exists because the database is being prepared for acquisition — the numbers have to survive a buyer's audit, not just a landing page.
SynergyDB is the intended foundation for the whole Synergy portfolio — the accounting ledger, the AI's long-term memory, the reporting store, and the event history are all designed to converge on this one engine.
Buyer-grade documentation (an eight-volume, ~1,250-page technical book), investor decks grounded in measured facts, and a public live demo where five database languages run side by side against one running copy.
Instead of lab benchmarks alone, the engine is proven by running GitLab, Odoo, OpenEMR, Mastodon, and the industry's standard order-processing stress test against it — each pass makes the engine stronger for every connection at once.
The wedge is consolidation for mid-sized companies and embedded use — one engine where teams run five today. Limits, rough edges, and roadmap items are documented plainly rather than rounded up.
Built in RustOne storage engineCrash-safe journalingAll-or-nothing updates~20 database languagesBuilt-in AI search3-server failoverEncrypted on disk