If a feature request doesn't align with the product's vision, it's perfectly fine saying no. If the vision itself is flawed, then the product is not likely to succeed.
Every vision is flawed, because we are fallible. The question is what we do about that. Do we keep the product nailed to the flawed vision until the last user has gone elsewhere, or do we use feedback from reality, from users telling us what they need, to improve the vision?