Capabilities

What we are actually good at

Not a service menu. Four delivery shapes that we have shipped repeatedly, with the constraints each one brings.

01

Custom enterprise software

Systems where the business rules are the product. We have built these for 15 years in environments where the accounting rules, the approval chain or the factory floor do not bend to a template.

Unusual business rules

Funding control, military price review, multi-level approval, branch performance scoring — logic that dissolves when you try to fit it into a generic ERP.

Legacy coexistence

Oracle / DB2 / SQL Server estates, SOAP interfaces, barcode and serial-number printers, SMS gateways, hardware serial ports. We integrate rather than replace by default.

On-premise and air-gapped

Offline installation, media control, role-based minimum privilege, audit trails. Common for defence, government and bank clients.

Source code handover

Delivered with source, schema, documentation and a runbook, so your team can take it over.

02

Platforms & multi-tenant SaaS

One platform base, several vertical products. Tenancy, RBAC, scheduled jobs, billing, observability and audit are built once and inherited by every product.

Multi-tenancy done properly

Tenant isolation, separate entry points, per-tenant configuration, unified operations.

Analytics that decides something

Forecasting and replenishment modelled on the real sales rhythm of e-commerce, not a generic time series.

Alerts that reach a person

Stock-out, dead-stock and warehouse-cost warnings routed by email to the person who can act on them.

Operable after launch

Gray release, metrics, cost attribution and a support model — the part that separates a platform from a prototype.

03

Hardware + software

Electronic systems, sensing devices and the software that drives them. Delivered with our technical partners; ownership and boundaries are agreed in writing before any client conversation.

Board repair and reverse engineering

Six-step closed loop: intake, analysis, localisation, repair, verification, archive. Includes re-designing obsolete boards rather than cloning them.

Automated board test

Test and fault diagnosis for digital boards and assemblies, with the software that records, computes and reports results.

Radar and sensing products

Radar level/velocity gauges, airport bird-strike detection radar, low-slow-small radar series, and the PPI console software behind them.

Edge to cloud

Device → edge → cloud → business application. Four layers, one contract, one maintenance path.

04

Applied AI engineering

AI that survives contact with operations: evaluated, cost-controlled, privately deployable, and wired into the workflow people already use.

Retrieval and agents

RAG over private document sets, agent workflows with tool calls, NL2SQL over business databases.

Speech and vision

ASR/TTS pipelines, image defect and content detection, recognition and classification.

Content pipelines

Article, video and MP4 generation lines that share a rendering and asset base — the work behind our own content factory.

Engineering discipline

Evaluation sets, regression tests for prompts, cost ceilings, fallback paths and private deployment.

Technology

Stack: we choose per client, not per fashion

Front endVue 3 · Element Plus · React 19 · ECharts · Tailwind · jQuery for legacy estates
Back endJava/Spring Boot · Spring (legacy) · FastAPI/Python · Node.js · .NET
DataOracle · MySQL · PostgreSQL · SQL Server · DB2 · Redis · Elasticsearch
IntegrationREST · SOAP · SMS gateways · barcode/packaging printers · SSO · hardware serial ports
DeploymentBare metal · VM · Docker · Nginx · offline/air-gapped · classified-network adaptation

Working principles

Four rules we do not bend

Reuse the client estate

We adopt the stack your team already operates unless a new technology earns its place and can be maintained after handover.

Written scope, then code

PRD and prototype before build. Change requests get an impact assessment before they get a sprint.

Daily build, weekly report

Progress proven by a runnable increment and a defect board, not by a status narrative.

Handover is part of delivery

Source, schema, docs, runbook, training. A system you cannot maintain is a system we did not finish.

Which of these is the problem you actually have?

Send a short description. We will answer with a point of view, not a brochure.