For some organisations, “just send it to the cloud” is a non-starter. Regulated finance, public-sector programs and anyone holding sensitive records need the whole system to run inside a boundary they control. We designed for that from the start rather than bolting it on later.
One product, three deployments
iDBQuery runs multi-tenant in the cloud, single-tenant inside your own VPC, or fully air-gapped on your own hardware — and it is the same product in each. Data residency is set per customer. The federation engine, the chat, the dashboards and the autonomous Analyst behave identically; only the boundary moves.
No outbound dependency
In the air-gapped configuration the AI and the agents run locally, with no call home. Your data is queried where it lives and never crosses the perimeter. Updates arrive through a controlled, audited channel rather than an open internet connection, so the security team keeps its guarantees intact.
Because the data is queried in place rather than copied into a shared store, the surface area you have to trust is small and explicit: the systems you already own, plus a tool that reads them. There is no third copy of your most sensitive records sitting somewhere you cannot see.
Sovereignty and capability are usually pitched as a trade-off — pick safety or pick power. Designing for the hardest deployment first is how we made that choice disappear: the air-gapped customer gets the same cited answers, the same live model and the same agents as everyone else.
Keep reading
Why cited answers beat a black-box chatbot
A confident number you cannot verify is worse than no number at all. We made “show your working” a hard guarantee — every figure traces to the exact source row it came from.
FEDERATIONOne live model vs. building a data warehouse
A warehouse is months of pipelines, schema design and a copy of your data that is stale the moment it lands. We connect your sources into one live, queryable model — no ETL, no second copy.
FINANCEPlain-language querying for finance teams
Variance, runway, cohort margins — answered from the actual ledger, in seconds, by someone who has never written a line of SQL. And every number is one click from its source.