Questions we hear
The questions teams ask before they hire us, with straight answers. Missing yours? Write to us.
Working together
Do you only build AI features? Mostly. We also build the software around them: the API, the admin screen, the dashboard and the deployment. We do not take projects with no AI part at all, because that is not where we are best.
Who will we work with? A small team led by the founder. The person who answers your first email designs the system and reviews the code, so nothing gets lost between the first call and the build.
Can you work with our own developers? Yes, and we like to. We can build the AI part while your team builds the rest, or we can pair with your developers so they can run the feature without us.
How long until we see something? A working first version usually takes about four weeks. You see the first working round, with its test results, before that.
How do you report progress? With a short written update every week and a call at the end of each round, where we show working software and its test results. The details are on the how we work page.
Which languages do you work in? English, for calls, documents and code. The features we build can answer users in other languages, and we test them in each language they must support.
Do you sign an NDA? Yes, before the first call if you want. Send us yours, or ask for ours.
What if AI is the wrong tool? Then we say so. Some problems are better solved with a search box, a rule or a form, and we will tell you before you pay for anything else.
Data and security
Where does our data go? Where you decide. We can keep processing inside the EU, use providers that do not train on your data, and sign a data processing agreement. We write down every place your data goes before we build.
Will our data be used to train a model? Not by us, and we choose model providers and settings that do not train on what you send them. If a provider’s terms change, we tell you.
Can the feature run on our own servers? Yes. We can use open models hosted in your own cloud account or on your own hardware, when the quality is good enough for the task. We measure that before you commit to it.
How do you handle personal data? We collect as little as the feature needs, remove personal data from prompts and logs where we can, and keep traces only as long as you allow. We follow the GDPR, and we document what is stored and why.
Can users see documents they should not? No. Search over your documents respects the same access rules as the documents themselves, and we test for it with cases where the right answer is “you cannot see this”.
What stops the feature from saying something harmful? Limits written into the design, filters on what goes in and what comes out, and tests that try to break them. Where a mistake would be costly, a person reviews the answer before it is used.
Models and cost
Which models do you use? The one that fits the job, its cost and your rules. We work with models from OpenAI and Anthropic, through Azure and AWS Bedrock, and with open models you can host yourself. We test more than one before we choose.
What happens when a better model comes out? We run it against your test set and compare quality, speed and cost with the model you use now. You switch only when the numbers say it is better.
What does a project cost? It depends on the scope. After the first call we send a written proposal with the price, so you know the number before any work starts. The first call is free.
What will the feature cost to run? We measure the cost of every answer from the first working round and give you a monthly forecast for your traffic. We also set limits per user and per feature, so a bad day cannot become a large bill.
Can you make it cheaper to run? Usually, yes. Shorter prompts, smaller models for easy questions, caching and better search all lower the cost. We change one thing at a time and check the test set after each change.
Code and ownership
Who owns the code? You do. The code, the prompts, the test sets and the documentation are yours, in your repository, from the first day.
Do you use AI to write the code? Yes, as a tool. Coding agents help us write and test code faster, and a person reviews every change before it is merged. We do not send your code or data to tools you have not approved.
Which technologies do you build with? Mostly TypeScript and Python, Node.js and Laravel on the server, React and Vue in the browser, and PostgreSQL with pgvector or a search engine for retrieval. We fit in with the stack you already have.
Will we be locked in to a provider? No more than you choose to be. We keep the model behind a thin layer of your own code, so changing provider is a measured change, not a rewrite.
Can you take over a feature someone else built? Yes. We start with a short review: we read the code, build a small test set and measure how the feature performs today. Then we tell you what we would keep and what we would change.
After launch
What happens after launch? We watch the traces, the costs and the quality reports closely for the first weeks, and fix what real use shows us. Then the feature is handed over, or we stay on a retainer.
What is a retainer? A monthly arrangement: a set amount of our time for new features, model and prompt updates, fixes and a monthly report on quality, speed and cost. We agree the amount and the fee up front.
How will we know if quality drops? The test set runs on every change, and alerts fire when quality, speed or cost moves past the limits we agreed. Feedback from users goes back into the test set, so the same mistake is caught next time.
Can our team change the prompts themselves? Yes. We show your developers how to change a prompt safely: change it, run the tests, compare the numbers, then release. Prompts are versioned like the rest of the code.
What if we want to stop? Then we stop, after a handover your team can run. Everything is already in your repository, so nothing is held back.
Still a question?
Write it down and send it. A real person reads it and answers within one working day.