Putting generative AI to work in your organization

This page covers the technical side of generative AI: which platforms to use, what types of projects are realistic, and how to secure them. For the human side — training your teams, governing usage, deploying an AI policy — see our AI activation service. The two often go hand in hand.

Claude, ChatGPT, Gemini: which platform for which use?

Platform	Strengths	Typical use cases
Claude (Anthropic)	Long-form writing, document analysis, agents with guardrails	Internal assistants, document automation, supervised agents
ChatGPT (OpenAI)	Versatility, large ecosystem of plugins/tools	Customer support, content generation, rapid prototyping
Gemini (Google)	Native integration with Google Workspace	Organizations already on Google Workspace

The right choice depends on your existing tools, the nature of the tasks, and your security requirements — it's one of the first things we evaluate together.

Types of projects we deliver

Internal assistants (RAG)

An assistant that answers your employees' questions based on your own documents and policies.

Document automation

Automated extraction, summarization, and filing of invoices, contracts, or incoming emails.

AI agents

Agents that carry out multi-step tasks with human supervision at key checkpoints.

Integration with Odoo / Microsoft

Connect generative AI directly to your ERP, CRM, or Microsoft 365 environment.

Security and compliance

Integrating generative AI raises legitimate questions: where is your data processed? Is it used to train the models? Does it comply with Law 25 on the protection of personal information? We build these questions into the design of every project from the start, not as an afterthought.

Assessment of data residency and processing
Configuration to exclude training on your data
Alignment with your Law 25 obligations

What about costs?

API costs related to generative AI (measured in tokens) can climb quickly without monitoring. Our FinOps service specifically addresses cost control: usage tracking, model selection, caching, and budgets.

See AI FinOps →

Frequently asked questions

It depends on your current tools and the type of tasks involved. We assess your context before recommending a platform — and it's possible to use more than one depending on the use case.

Not by default on the enterprise offerings of major providers, but the configuration must be verified and documented. This is a systematic step in our deployments.

Yes, this is one of the most common types of projects: an agent or assistant that interacts directly with your ERP or CRM data via API.

Often with an AI activation component: training teams and identifying 1-2 simple use cases before investing in more advanced technical integration.

Have a generative AI project in mind?

Let's discuss your data, your tools, and your security constraints.

Book a discovery call