Skip to content
Neorimia

Process

AI-assisted development, with human review at every release.

AI tools can help produce code, investigate problems and draft tests. A working business application also needs clear requirements, architecture, review and careful release decisions. Gabriel is responsible for those decisions throughout your project.

The steps

From problem to handover

  1. 1. Agree the problem and the acceptance criteria

    We identify the users, critical workflows, integrations and constraints. The proposal states what will be delivered and how you will check that it works. For an AI feature, we also agree how output quality and exceptions will be evaluated.

  2. 2. Define access and data handling

    We agree which tools and providers may be used. Development uses synthetic or redacted examples where possible. Production access is limited to what the work requires. Personal or confidential information is not sent to an AI provider without reviewing the data handling requirements and agreeing the permitted use.

  3. 3. Divide the implementation into reviewable changes

    Gabriel defines the architecture and breaks the project into tasks. AI agents can work on bounded implementation, test or documentation tasks. Parallel work is used where tasks are independent; the number of agents does not determine your delivery commitment.

  4. 4. Review the code and the behaviour

    Generated changes are checked before they are accepted. Tests cover the agreed critical journeys, including relevant failure cases. Gabriel reviews the implementation and the test expectations. Authentication, permissions, data boundaries and external integrations receive explicit checks when they are part of the system.

  5. 5. Validate AI features separately

    Where the application itself uses AI, we evaluate it on agreed examples, record failure cases and define when a person must review the output. Sensitive actions can require approval before execution. We agree monitoring, usage limits and a way to stop or bypass the AI workflow.

  6. 6. Review progress and approve release

    You review regular demonstrations in a test environment. We track decisions and separate corrections from requests for new functionality. Production release follows the agreed acceptance process and a deployment plan with a recovery path.

  7. 7. Receive the code and operating notes

    You receive repository access, installation and deployment notes, and the agreed account access. Rights to custom deliverables transfer under the contract after payment. Open-source components and third-party services keep their own licences and terms.

  8. 8. Know what happens after launch

    For 30 days after launch, defects in the agreed implementation reported during that period are corrected without an additional development fee. New features, changes requested after the design has been approved and subsequent third-party changes are outside that commitment. Ongoing maintenance is optional and scoped separately.

Transparency

A process you can question

Ask to see the scope, acceptance criteria, data handling choices and handover plan. AI is part of how we work; it does not remove our responsibility for what we deliver. Security certifications, independent audits and round-the-clock support are separate requirements and are not implied by this process.

Gabriel Cruceanu leads the project and remains your direct technical contact. AI agents assist with selected tasks. Additional specialists are involved only when their role and cost have been agreed.

Yes. AI can assist with implementation, tests and documentation. Gabriel reviews the changes and remains responsible for architecture, integration and release decisions.

The contract defines the transfer of rights to custom deliverables after payment. You receive the agreed source code and documentation. Open-source software and third-party services retain their own terms.

We agree the permitted tools and data handling before work starts. We use synthetic or redacted examples where possible and review the requirements before sending personal or confidential data to an AI provider.

The handover includes source code access, account access and operating notes so the project can be continued by another engineer. If your business requires guaranteed backup staffing or emergency response, we must agree that arrangement before accepting the project.