Skip to content
TEN INVENT S.R.L.

How we use artificial intelligence

We build AI-powered products and use AI daily in our work. This page explains what tools we use, how we integrate AI into client deliverables, and our stance on the EU AI Act.

Last updated: April 2026

01 — Internal use

Tools we use in our daily work

We actively use AI assistants and tools to write better code, think through problems faster, and produce higher-quality work for our clients. We don't hide this — it's part of how we operate.

Claude (Anthropic)
Architecture decisions, code review, documentation, writing.
ChatGPT (OpenAI)
Research, drafts, exploratory problem-solving.
AWS Bedrock
AI infrastructure for client projects — model APIs, RAG pipelines.
Cursor / GitHub Copilot
AI-assisted coding, autocomplete, inline suggestions.
LM Studio (local)
Local inference on private data — nothing leaves the machine.
Strands Agents (AWS)
Agent orchestration in production systems we build.

Every AI-generated output — code, text, or analysis — is reviewed by a human before it goes anywhere near a client or a production environment. We don't ship what we don't understand.

02 — Client projects

AI in the products we build

When we build AI-powered systems for clients, we act as a technical development partner. The client owns the product and is responsible for how it's deployed and used. Our job is to build it well.

We build to specification. Compliance with applicable regulations — including EU AI Act, GDPR, and sector-specific requirements — is the responsibility of the product owner. We are happy to advise and to build systems that make compliance easier, but we are not the deployer.

In practice, the systems we build typically include:

Human-in-the-loop checkpoints Audit logging Transparent AI responses Configurable safety guardrails Source attribution in RAG systems Role-based access controls
03 — Our principles

How we think about AI

  1. 01

    Humans decide

    AI assists, summarizes, and accelerates. It doesn't make final calls on things that matter — people do. We build systems that keep humans in control.

  2. 02

    Transparency by default

    If an interface uses AI, users should know. We don't mask AI interactions as human ones. We label, log, and explain.

  3. 03

    No client data in public models

    We don't feed client or user data into third-party AI models without explicit agreement. Sensitive workloads run on private infrastructure or local models.

  4. 04

    We stay current

    AI regulation is moving fast. We track EU AI Act developments and update our practices as the landscape evolves.

04 — Regulatory stance

EU AI Act & compliance

The EU AI Act (Regulation 2024/1689) broadly applies from August 2026, with remaining obligations phasing in through 2027. We've assessed our position:

As a software development consultancy, TEN INVENT acts as a technical subcontractor. We are not the deployer of the AI systems we build for clients. Responsibility for AI Act compliance of the deployed product rests with the product owner.

For our own internal use of AI tools, we apply the following:

›

AI literacy

All team members working with AI tools maintain documented awareness of their capabilities, limitations, and appropriate use — as required by Article 4 of the AI Act.

›

Inventory & classification

We maintain an internal inventory of AI systems used and have classified them by risk level. The tools we use internally fall in the minimal to limited risk categories.

›

No high-risk deployments (internal)

We do not use AI internally for recruitment decisions, employee evaluation, or any process that would qualify as high-risk under Annex III of the AI Act.

Questions about how we use AI in a specific project? Contact us at ion.anghel@teninvent.ro

TEN INVENT S.R.L. · CUI 52212790 · contact@teninvent.ro

This page is updated as our tools and practices evolve.