AI Integration — Frequently Asked Questions
Last updated: September 27, 2026
Frequently Asked Questions
How long does an AI integration take?
We don't quote a fixed timeline, because the workload and scope differ for every project. We determine the timeline in the first conversation, once we understand the work, and present it to you as a concrete schedule. An estimate given before the scope is clear tends to become a promise nobody can keep.
How is the price of an AI project determined?
We don't have a price list. Pricing is set per project once the scope is clear, because two projects under the same heading can be very different. We start by understanding the need, define the scope together, and the price follows from that.
Will we own the source code of what you build?
Yes. In every service and product we build, the source code belongs to you — down to the last line. Our aim is for each client to own a product that is genuinely theirs, not to hold the code and keep you dependent on us.
Do you provide support after delivery?
Every digital product or service we deliver comes with 1 month of free support; usage questions and any issues you run into are covered during that period. After that, we set up a separate support agreement if you need one. Beyond that, our defect warranty has no time limit: if a fault originating from us appears in what we delivered, we fix it no matter how much time has passed. We don't hand over work that doesn't function. Everything else is treated as separate work: new feature requests, scope extensions, and adaptations required by changes third parties make to their own systems (operating systems, payment providers, APIs).
Which AI model do you use?
The model is chosen per project, and we don't tie ourselves to a single provider. We write the setup so the provider is isolated in one place; when the model or provider changes, the rest of the application doesn't. That matters because prices and limits on the AI side change often — if a switch becomes necessary one day, it shouldn't require a rewrite.
What happens if the AI gives wrong information?
This is the core question: what the assistant knows and doesn't know is decided by its source. We limit answers to content you have verified, and when a question isn't covered by that source, the assistant says so and points to the relevant page or to contacting you, rather than inventing an answer. A wrong answer costs more than no answer.
Will our data be used to train the AI model?
That depends entirely on which provider you work with and under what terms; it's one of the first things to settle when choosing a provider, and we clarify it before setup. What we control on our side is that the boundaries of what the assistant is given stay in your hands.
Can you train a custom AI model for us?
No — model training and fine-tuning are not our area. What we do is connect capable existing models to your system and your content, correctly. We say this upfront, because "custom AI" is often taken to mean model training, while in practice what's needed is usually a properly built integration.
Can AI be added to our existing software?
Yes — this is the work we do most often: adding an AI-powered feature to a running website, platform or admin panel. There's no need to rewrite the system. The assistant on our own site works exactly this way: the site came first, and the assistant was added later.