2:22
Privacy by design for AI: what should actually change?
What should privacy by design make an engineer do differently? Anthony explains seven principles, practical data boundaries and the evidence users should expect.
Watch on YouTube · Join the conversation
Anthony narration supplied by Mevanu’s Founder and generated with ElevenLabs. Original explanatory cards use a fictional meeting-notes-to-proposal example; they are not app footage. Brava remains in development. The seven principles are design and launch requirements, not a claim of verified fulfilment. Vivaldi Spring I, Michelle Ross, Odyssey Orchestra and Eric Jacobsen, licensed through Epidemic Sound.
Read the transcript
Text follows the published English captions, including the original narration’s wording.
English captions (SRT)English captions (VTT)
Hi, I’m Anthony from Mevanu.
“Privacy by design.” It sounds reassuring. But what should those words make an engineer do differently? And how can you tell whether you’re getting meaningful protection or a well-written marketing line?
At Mevanu, privacy by design means building protection and human control into every stage of a task—and making those decisions open to scrutiny.
Here’s what that asks us to build.
Imagine asking an AI assistant to turn meeting notes into a client proposal. Useful work. Possibly sensitive work, too.
First, anticipate the risk. What could go wrong if those notes reached the wrong person? Address that before it happens.
Choose protective defaults. The task needs relevant notes. Why should your entire inbox come along for the ride?
Build privacy into the architecture: what the assistant can read, what it can share, and when it must ask. Those boundaries need to shape how the software works.
Preserve the useful part. Could the proposal work with fewer personal details or narrower access? Good design should help you get the job done while protecting your information.
Protect the whole journey. Where are the notes stored? Who can access them? How long are copies kept? Plan for deletion, too.
Make the practices checkable. Explain what happens to the data. Give people a way to question it. Allow independent scrutiny.
And keep people in charge. Clear choices. Understandable permissions. A way to say “stop.”
Those are the seven principles guiding Brava’s design. Brava is in development, so we still have to demonstrate how well we meet them.
That means connecting each principle to an actual safeguard, checking whether it works, and being clear about what remains unresolved.
When a company promises privacy by design, ask it to show that connection. What information does this feature need? What happens when you refuse permission? Who has checked the answer?
Ask us, too.
For the full guide, visit Founder’s Bookshelf at mevanu.ai. It expands these principles with examples, questions to ask and a review note you can use.
Have an idea? Drop a prompt, comment, suggestion, or question.
Mevanu. Just. Smarter.
Want help defining sensible data boundaries and review questions for your AI workflow? Talk with Mevanu.