SynapCores v1.14.1 — Bring your MongoDB data in

Your operational data rarely starts life in the database you want to query it from. It's in MongoDB, in Postgres, in Snowflake — and getting it somewhere you can join it against vectors, a knowledge graph, and in-database LLMs usually means a pipeline you have to build and babysit.

v1.14.1 shortens that. SynapCores' data-sync engine now speaks MongoDB, and data sync is on by default.

A collection becomes a table

Point SynapCores at a MongoDB collection and it replicates into a normal, queryable SQL table. Documents map the way you'd hope: _id becomes the primary key, and nested fields land in a JSON column you can reach into with JSON_EXTRACT.

SELECT _id,
       JSON_EXTRACT(doc, '$.customer.name') AS customer,
       JSON_EXTRACT(doc, '$.total')         AS total
FROM   orders_synced
WHERE  JSON_EXTRACT(doc, '$.status') = 'shipped';

Once it's a table, it's just SynapCores — embed it, graph it, hand it to an AGENT_RUN workflow, join it against everything else you already keep here. Data-sync sources now cover MySQL, PostgreSQL, Snowflake, and MongoDB, with SynapCores as the target.

Data sync used to be off unless you flipped an environment variable — which meant the UI showed connectors that answered "data sync not configured." Now it just works out of the box. (Prefer it off? AIDB_ENABLE_DATA_SYNC=false.)

Your graph got ~20× faster, too

While wiring data-sync-on-by-default, we found — and fixed — a graph-storage bug. Graph queries had been re-opening a per-tenant storage handle on every single request. With a background scheduler now always running, that occasionally raced itself and could take the whole gateway down. Graph storage is now opened once and shared, the same way the rest of the engine works.

The stability fix came with a speed bonus: graph query latency dropped from ~38 ms to ~2 ms at the median — roughly 20× — and under concurrency the graph went from dropping most requests to serving 1,700+ per second on our test box. Your SQL paths are untouched.

And the console caught up

The web console now matches what the engine actually does: the Data Sync page is wired to the live backend — create and test connections, run jobs, watch real metrics, add a MongoDB source — instead of a mockup. Native in-process vision from v1.14.0 finally has a one-click "Describe (AI vision)" action (and the UI no longer implies you need a cloud key for it). The in-app SQL reference and the "what's in Community Edition" page were brought current.

Validated

feature_validator 88/88 · recipe certification 158/162 with zero regressions vs the prior release · zero gateway aborts across repeated data-sync-on certification sweeps · non-AVX-512 canary boots and runs the models with no illegal-instruction · verified on the published Docker image.

Get it

docker pull synapcores/community:v1.14.1-ce
# or the one-liner:
curl -fsSL https://get.synapcores.com | sh

One note on scope: real-time MongoDB CDC (change streams) is an Enterprise feature. In Community Edition, MongoDB comes in through batch data-sync — which, for getting your collections queryable next to everything else, is exactly the point.