FAQ

With a thirty-minute call. We start from your business, not from the technology: the repetitive tasks, the documents piling up, the decisions made without the right information. From there, together we pick a first use case that is useful, realistic and measurable, and take it to production quickly.

It isn’t always, and we will tell you when it isn’t. Some needs are met with simple rules, a good dashboard or conventional automation. We only propose a model or an LLM when it delivers something a simpler solution cannot. That is what scoping is for: deciding before investing.

No. Most of our clients start with documents, emails, exports from business tools and Excel files. We assess what your data can support today and what needs consolidating, and we build with what you have. We can be your data team, or work with yours.

It depends on the engagement model. An embedded engagement is billed by the day. A fixed-price MVP starts at €5,000 excl. VAT, with the scope, price and timeline set before we begin. A larger project is scoped, then estimated phase by phase. In every case, you know the cost before we start.

Weeks, not months. We work in short increments: a first useful version goes live quickly, then grows based on real use. A larger project is split into phases, and each one delivers something that works.

Directly with the team that does the work, with no middlemen. At Airbay Data, the person you meet on the first call is the one who scopes the project, builds it and puts it into production. Close to ten years in the field, five of them with Airbay Data, and more than thirty projects for companies of every size, from SMEs to large enterprises.

Both. We are based in Paris and work on site in the Paris area, remotely everywhere else. The pace, the on-site days and the team meetings are agreed with you at the start.

Yes. Your data remains yours and is never used to train third-party models. We work in your environment, under your access rules and in compliance with the GDPR, and choose models to match the confidentiality required: proprietary models from the major providers, or open-source models deployed in your own infrastructure or with a European hosting provider. We are happy to sign a non-disclosure agreement from the very first conversation.

No. We build on what you already use: your cloud, your databases, your business tools. We only suggest a new tool when it brings a clear benefit, and never one that locks you in to us.

You do. The code, the models, the data and the documentation are delivered into your repositories and your accounts, and all access is handed over to you at the end. What we build keeps running without us.

We measure it before go-live, on your real cases: your teams’ actual questions for an assistant, your documents for an extraction, your historical data for a forecast. We define the metrics with you, in business terms, and track them once in production.

An AI system is never finished: the data changes, and so does the way it is used. Every delivery includes basic monitoring and a handover to your teams, with documentation and training. From then on, you can run it yourselves. If you wish, we stay on for support and further improvements, embedded or phase by phase.