Samvrik: frequently asked questions

Samvrik is a sovereign data and operations platform by RUSH. It acquires the external data your operations depend on, harmonizes it with your own data into one governed record, and runs governed operations and AI on top of that record, all inside your own environment. It is made of three platforms: Quarry, Smelter and Assembler.

Quarry acquires external data, such as public registries, tenders, prices and availability, and keeps it current. Smelter performs data harmonization: it brings every internal and external source into one trusted, governed record. Assembler acts on that record, running your business rules, workflows, applications and AI agents with every action audited. Together: acquire, harmonize, act.

No. Each platform removes one bottleneck on its own and can be adopted independently. Most organisations start with the platform that addresses their most pressing problem and add the others when they earn their place. When combined, they carry an external fact all the way to a governed, audited action.

Samvrik is built for organisations that need to own what they run: regulated industries that must stay on premises, organisations with data-residency obligations, and growing enterprises that want enterprise-grade data and AI capability without the cost or lock-in of the largest platforms. It is used in sectors such as trade and logistics, travel, banking, corporate finance and the public sector.

Samvrik comes from the Sanskrit sam (shared) and vṛtta (the record of what occurred): the shared record. That is the idea at the centre of the product: one trusted record of what happens in your business, which you own.

Inside your environment only: your private cloud or your own data centre, and air-gapped where required. RUSH does not offer Samvrik as a hosted service and holds no copy of your data.

Your choice. Your own team can run it, or RUSH can operate the stack for you inside your walls as a managed service. Either way the infrastructure, the data and the record remain yours.

Yes, the platforms are designed to run with no connection to the internet, which is common in defence, banking and the public sector. The one natural exception is Quarry's acquisition itself: Quarry reaches only the external sources you choose to collect from, and nothing else. AI features can run entirely on models you host yourself.

Samvrik runs on standard containers, so it deploys on any major private or public cloud account you own, on your own servers, or on Kubernetes. It does not depend on any one cloud provider's proprietary services.

You do. Every byte stays inside your perimeter. RUSH has no data-monetisation business and sells no competing applications, so there is no competing agenda over what flows through your record.

Access is enforced by the platform, not left to each application. Users sign in through a single identity layer, and permissions apply down to the individual record and field. Each workspace is isolated from every other at the data layer, and every action is written to an audit trail.

Because the platforms run entirely inside your environment, data and its processing, including AI inference if you choose self-hosted models, stay where your regulations require: for example under India's DPDP Act, EU data-residency rules or GCC requirements. Samvrik provides the controls, such as access control, masking of sensitive fields and audit trails; compliance remains an assessment of your overall deployment.

In Assembler, every AI call passes through the same governance pipeline: access control, guardrails, budget checks, tracing and audit. AI agents act only within the rules and permissions you define, high-impact decisions can require a human approval step, and every model call is recorded. You choose which model serves each capability, including models hosted inside your walls.

Quarry acquires government, licensed and public-origin sources, plus targeted extraction that you deploy and operate in your own infrastructure. Each source's terms of access are assessed per engagement, and the platform is configured to respect them, including per-source rate limits.

No. Samvrik orchestrates widely used open-source components and keeps your data in open formats, such as PostgreSQL and Apache Iceberg tables, that any compatible tool can read. Your team can extend it with standard languages and tooling, and there is no proprietary language to learn.

Samvrik builds on proven open-source foundations, including PostgreSQL, Apache Kafka, Temporal, Apache Iceberg, Trino, dbt, Meltano and the Ory identity stack, and adds the integration, governance, data-quality and AI layers that turn them into one platform. The value is in how they are orchestrated and governed together.

Yes. For each component you can choose the community edition or, where one exists, an OEM-supported self-managed edition procured through RUSH, on the same architecture. RUSH is your single point of contact for support and escalates to the component vendor where needed. Licences can move to your name, or back to open source, without re-platforming.

Through standard interfaces: REST APIs, events, webhooks, database and file connectors, and connectors for business systems such as ERP and CRM. Each platform is deployed independently and they connect to each other the same way, so you can adopt them one at a time.

Each platform is licensed for a fixed term of 3, 5 or 10 years, in three editions: Starter, Growth and Enterprise. Each edition includes a defined allowance, such as capacity, users, sources or applications depending on the platform, with add-ons available above it. Annual support and updates accompany the licence, and managed operations by RUSH are optional. Pricing is scoped per engagement; talk to us for a quote.

A term licence gives you a fixed, predictable cost for the life of the term, for software that runs in your own environment under your control. Longer terms lower the effective yearly cost.

Book a scoping session. In about 45 minutes with the people who built it, we map where Samvrik removes the most from your roadmap and which single platform to pilot first. Every engagement starts with a time-boxed pilot inside your own environment, on your own data.

Scope, deploy, model, operate, expand. We scope the first use case, install the platform in your environment, model your data, rules, workflows and sources, then run it on your live data. You add the next platform where the next constraint is, and the record compounds in value as it accumulates.

Onboarding (installation, single sign-on and administrator handover), annual support and updates, and optional services: building applications, sources or data models for you, and running the platform for you inside your walls.