Your conversations' privacy is essential to the trust you have in using Ava, and this is why we go above what's been done in the industry to respect it.
First, we use your voice and conversation data to teach Ava. This is how artificial intelligence learns how to get better to:
- Recognize your voice (or your peers) among all the rest to distinguish better who speaks in conversations.
- Transcribe what you (or your peers) say in conversations.
What are our economic incentives as a social entreprise?
Ava's economic model is to deliver an accessibility service, which is cross-subsidized: organizations and heavy users pay a monthly subscription rate for Ava plans. Part of this revenue covers their usage of Ava and additional premium features, and the rest enables us to fund 5 free caption hours every month for the rest of Ava users.
There is no other revenue source, hence Ava economic model is completely independent from advertising dollars. We have no economic interest to sell your data or allow any 3rd party to access it, and intend to keep it this way to offer a sustainable economic model to affordable & total accessibility.
Your conversation data is secure, anonymized and fragmented: we remember how people sound, not what they said.
- We use enterprise-grade storage of your data, and encrypted end-to-end protocols to broadcast the audio. As a result, we pass security tests of top Fortune 500 companies, including banks.
- When an Ava conversation happens, the audio is fragmented in "sounds", and we use these to learn how people sound when they speak. We don't capture or store what the conversation is about.
- Because the conversation data is anonymized, no one can access it unless we have your written authorization.
- If you choose not to save the conversation transcript, you cannot access the transcript later because it is not kept in Ava. If you save the conversation notes, they are text-only and cannot be used as evidence or proof because they can be easily modified.
Why isn't Ava data local?
Ava service is based on a powerful speech-to-text algorithm. In order to deliver the highest accuracy possible for your conversations, Ava needs a high computation power that smartphones cannot deliver. Hence, using Internet networks is more beneficial to deaf & hard-of-hearing inclusion.