AITEH
RO
Case studies
Confidential client

Confidential client

Sugar production

AITEH NexFlow

Five websites managed in one place and synchronised with SAP

About the client

One of Europe’s largest sugar producers.

We cannot publish this client’s name for confidentiality reasons. We describe only what we built.

5

websites managed in one system

What we built

  • Centralised management of websites across multiple countries, with a dedicated website for each country.
  • Editors from every country, each publishing content for the countries they are assigned to.
  • SAP integration to synchronise data.
  • Management of digital content and structure.
  • The same content also feeds the mobile app.
  • Streamlined workflows between websites and ERP.

How we worked

We contributed to developing the system as part of the project team.

The result

Content is managed consistently, and data remains aligned between the websites, the mobile app and ERP.

Controlled delivery to every environment: separate CI/CD and a review before deployment

The challenge

  • A single CI fed three environments (DEV, QAS, Prod). Deployments sometimes started for the wrong environment or directly on push, without waiting for the build, and could redeploy an old build.
  • Some checks did not actually run: code analysis had never been executed, and the migration step in the deployment reported success without running anything.
  • An automatic dependency “fix” had silently changed major library versions and broken the build.
  • Linting and unit tests were disabled, so problems were only discovered after merging.

What we built

  • Separate CI and CD for each environment, each tied to its own branch (develop → DEV, main → QAS, production → Prod). A deployment starts only after its own environment’s CI; production requires manual approval.
  • Shared templates: stages are defined once, and environments differ only by parameters, so the pipelines cannot drift apart.
  • Deployment with no installation on the server: CI builds the application and includes the production dependencies in the artefact, so the server installs nothing at deployment. Migrations run once, when the application starts.
  • Parallel checks: linting (including SonarSource quality rules run locally, without the code leaving the build agent), unit tests with coverage, the build and an npm security audit.
  • A single go/no-go review: if anything is found, the pipeline stops and shows a one-line summary for each check (lint errors and the main rules, failed tests, packages with high/critical vulnerabilities). The run page holds the full report: every failed test with its error, every vulnerable package with its available fix and the complete lint report. “Go” deploys, “No-go” stops the release; with no answer within 24 hours, it counts as “No-go”.

The result

  • Deployments reach only the correct environment, from the correct build.
  • Every release is a conscious decision, with all problems visible in one place.
  • Roughly 2.5 minutes saved on every CI run by replacing cloud code analysis with local rules.
  • Hidden problems identified and fixed: the migration step that did nothing, the accidental major version changes and the tests that were not running.
  • More than 1,300 unit tests and a full lint report run on every build, including on feature branches.

Wondering what could change in your business?

We offer a free 25-minute review of your business. You leave with a clear list of what you could automate.

Book your review

More case studies

Oana Teșoi

Oana Teșoi

Customers pay during the call, invoices issue themselves, and AI helps create email campaigns

Read the case study