Embedded like DuckDB. Multi-tenant on object storage.
Built for the AI of this decade.
Faster than Kùzu on every measured benchmark.
In 2024, S3 quietly shipped conditional writes. The last missing primitive. For the first time, you can build a coordinated, durable system where object storage is the database — no Raft, no ZooKeeper, no etcd. The recipe has paid off for vectors, for queues, for analytics. It has not been done for graphs.
In October 2025, Apple acquired Kùzu and archived the repository. The most thoughtful columnar graph engine ever published went quiet. A hole opened.
Vector search is necessary. It is not sufficient. The systems being shipped — assistants, copilots, deep retrieval engines — all need graph primitives at scale. The database for that decade has not been built.
So we are building it.
And the engine is real.
Embedded
Like SQLite for relationships. A single dependency. No daemon. No cluster. Your graph lives in a file on your laptop, in your container, or on S3 — your choice.
$ pip install namidb
Server
Point it at an object store. Done. No volumes to provision. No state to migrate. Restart anywhere — the data was never on the box to begin with.
Cloud
Namespace per agent. Namespace per tenant. Scale to zero when idle. Pay for what you store, not for headroom you don't use.
The same engine across all three. Move when you're ready. Never rewrite.
Every assistant worth shipping needs three things: a way to remember, a way to retrieve, and a way to reason across what it found. NamiDB ships with the primitives.
Hybrid vector + graph retrieval in one engine — no federated joins, no glue code. The substrate for GraphRAG that doesn't add a second store to your stack.
Snapshot isolation across time. Reconstruct what an agent knew on any past date — by design, not by hack. Namespace-per-agent in the cloud.
The two standard query languages for graphs, in one engine. Cypher today. GQL — the ISO 39075 standard ratified in 2024 — next.
Natural language in. Validated query out. Schema-aware. Cost-aware. Designed into the engine, not bolted on top.
Kùzu was the columnar graph engine that defined the category — until Apple acquired it in October 2025 and the repository went quiet. We took the queries it was designed for, ran them on our engine on identical hardware, and measured what came out.
Recent messages from your friends
22× faster than Kùzu.
Recent likers of your messages
1.4× faster than Kùzu.
Recent replies to your messages
55× faster than Kùzu.
Friends of friends, with their messages
3.5× faster than Kùzu.
LDBC SNB Interactive on the synthetic SF1 datagen. object_store::InMemory. Same hardware as the Kùzu run. Compared against Kùzu 0.11 — the immediately-pre-archive release. SF10 separately validated: all four queries pass the 2× gate. Methodology and raw runs live in docs.namidb.com/bench.
Four of four queries pass the gate on both scales. NamiDB is faster than Kùzu on seven of eight (query, scale) cells. The day a number flips against us, this page will say so.
Legacy graph databases price by RAM. Double your hot set, double your bill — even if your data didn't grow. NamiDB stores in object storage at object-storage prices. Compute is yours to size. Storage is what you actually use.
Up to 100× less expensive than managed graph databases priced by RAM.
Estimate derived from object-storage pricing versus the RAM tiers of managed graph databases at equivalent dataset sizes. Measured per-tenant costs publish when our cloud enters public beta.
Every architectural decision lives in an RFC. Every benchmark is published — including the ones that don't go our way. The engine is source-available from day one under BSL 1.1, and each release converts to Apache 2.0 three years later. A commercial license is available for teams that redistribute it as a managed service or embed it in closed-source products.
Fonles Studios is a research and product studio with one obsession: the substrates the next decade will run on. NamiDB is built by a multidisciplinary team — engineering, research, legal, operations — coordinated by the four cofounders below. Our third project. The first two — Alhelí and Tarovers — taught us why.
Founder & CEO, Head of Engineering
Built his first software at eleven. Became the youngest CTO in the history of an international cybersecurity firm. Founded Alhelí (legal AI) and Tarovers (behavioral simulation) before NamiDB. Now leads Fonles Studios and the engineering team behind every release.
Co-founder, Head of Legal
Heads legal. The contracts, the IP, the licensing terms — the scaffolding that makes shipping at this pace possible.
Co-founder, Head of Research
Academic of fifteen years with hundreds of peer-reviewed publications. Leads the NamiDB research team — sets the bar for what we cite, what we measure, and what we publish.
Co-founder, Head of Operations
Heads operations. From the cap table to the deployment cadence — the workflows that keep the team shipping on schedule.
With engineers, researchers, designers, and operators behind every release. Fonles Studios, Corp. — Delaware, USA.
We are opening access in waves. The first wave is for engineers and teams who want to shape what NamiDB becomes — through early use, candid feedback, and design partnership.
We're giving it a database
worthy of the decade ahead.