Your data model. The right engine.

Open-core data infrastructure, built around your workload.

We are building Smart Data Engine around your declared model and constraints. A paid control plane plans placement; open SDKs connect directly to your engines, on your infrastructure or hosted by us. When standard engines fall short, we can build a specialised one.

Discuss your workload Explore the product

Open client libraries. A public database engine. Code and benchmark methods you can inspect before we talk. Explore the evidence

Measure the workload. Then choose the engine.

A transactional database, an analytics engine and a purpose-built store solve different problems. We start with your access patterns, constraints and a measurable baseline. Together, we agree the performance and resource targets, then verify the improvement against that baseline.

Read Our Approach

Three Ways to Work With Us

A data layer, a database we run for you, or a purpose-built engine. Choose what your workload needs.

Managed

Smart Data Engine

Your logical model. A physical layout that can evolve.

Our early-stage flagship combines open SDKs with a paid control plane. Declare entities and constraints; the SDK connects directly to the selected engines. We agree the workload, integration and rollout scope before production use.

  • AI proposals checked against your constraints
  • Locally verified migrations with an agreed rollout
  • Keep using the last placement after the subscription ends
Explore Smart Data Engine
Standalone

Database Hosting

Your application. A database we operate.

Plan a managed PostgreSQL or ClickHouse database with us. Your application connects directly using standard database clients. Capacity, data location and operations are agreed before production use.

  • No SDE or SDK required
  • Planned billing by consumed SDE Compute
  • Data, backup and transfer limits agreed up front

Available by agreement. Self-service provisioning and compute billing are not yet available in the portal.

Discuss database hosting
Built From Scratch

Crafted Database Engine

One workload, one machine, no generalisation

When tuning runs out of room, the engine itself is the problem. We write a new one in C++20 around your access patterns and your hardware, down to the storage device.

  • Latency budget agreed before we start
  • Resource targets verified on your workload
  • Designed for your hardware, down to the SSD
Explore a custom engine

Judge Us by Our Code

Public source, reproducible methods and explicit limitations.

Our L2 orderbook engine is written in C++20 for one workload: keeping every level of every book and answering what the book was at any moment. Its source, tests and benchmark methods are public.

  • C++20 Native L2 orderbook engine
  • Open Source under Apache-2.0
  • Measured Public benchmark methods

Results depend on hardware and workload. The public comparison documents its methods and limitations; it is not a performance guarantee for your application.

View the source on GitHub

What it does and what we measuredRead benchmark methods and resultsInspect the open client libraries

The technology behind the work

SDE adapters are available for PostgreSQL, ClickHouse and our orderbook engine. Capabilities differ by library and engine; check the current support matrix before choosing a workload.

Current SDE engines

PostgreSQL ClickHouse Orderbook DB Engine

Custom-crafted for financial order books, with a focus on low latency and low cost.

AI integration

Your model endpoint Capability checks Audited proposals

Infrastructure

AWS Azure Google Cloud OVH On Premises

Engineering languages

C++20 Python TypeScript

Check current SDK and engine support