do you use AI to write the code?
we use it the way any current team does: scaffolding, tests, review. a human still reads every line before it ships. what we don't do is bolt an AI feature onto your product because it's fashionable.
AI tooling is part of how code gets written now, and pretending otherwise would be odd. we use it where it saves time without changing the outcome: the boilerplate of a new module, a first draft of the tests, a second pair of eyes during review. it's a faster keyboard, and the judgement stays with a person.
what doesn't change
a human reads every line before it ships, and that person answers for it. generated code goes through the same review, the same tests, and the same pre-launch security pass as anything else: authentication, access control, injection, secrets, headers, dependencies. if you want to see what that pass covers, the checklist is public.
the tool choice doesn't change either. we still build on boring technology another developer can pick up cold, because the code has to outlive whichever assistant was fashionable the month we wrote it.
AI features in your product
that's a separate question. we build AI features when there's a real use case: a task that costs your team hours and that a model handles reliably. we don't add one because a competitor did, and the rule is the same as for any new service: it has to remove more work than it adds. if you're not sure whether your idea qualifies, describe the task and we'll tell you.