Evro AI - Trust Center
OUR COMMITMENT TO YOUR PRIVACY

Your communication data is safe. Here's exactly how.

Complete transparency on how we protect your data, the science that informs our platform, and what people who shaped what we built say about it.

We built Evro with one non-negotiable: don’t build features that we wouldn’t build for our own friends or family. We hold ourselves to that standard because we believe you deserve the same care and protection as the people closest to us.

How we protect your data.

We built Evro to be a trusted partner in learning better ways to communicate. That comes with real responsibility. Here's what we do with that data, and what we will never do.

Your data never trains models

We default to zero-training on customer data, with strict isolation across all users and workloads.

No hidden third-party data sharing

We disclose every subprocessor and never pass your data beyond audited, contract-bound partners.

Agents act only with permission

Every action is scoped, logged, and revocable, so nothing happens without your explicit control.

Delete means fully deleted

We permanently remove your data from live systems, backups, and embeddings on request.

Built for secure work usage

Enterprise controls prevent unauthorized tools, keeping sensitive data inside approved environments.

Anonymity stays truly anonymous

We minimize data collection and prevent re-identification through strict separation and aggregation.

Protected against prompt injection

We detect and block malicious instructions to stop unauthorized data access or exfiltration.

Your personal data stays private

Sensitive conversations are encrypted and never used to build behavioral or emotional profiles.

Real professionals who shaped this product.

Evro was built with the people who needed it most. Our beta researchers were professionals navigating high-stakes meetings, communication challenges, and the gaps that no existing tool was filling. They helped shape decisions we made.

Isobels
Isobels
Jacob
Jacob
Maria
Maria
Simone
Simone
Tariq
Tariq

Verified customer reviews.

We aren’t live for purchases or independent reviews. Once we are, Evro AI cannot edit, filter or remove these reviews. Every reviewer is a verified purchaser.

★★★★★
Invitations sent to every customer equally after delivery. We responded publicly to all feedback. Our Trustpilot rating updates in real time.

The science behind every feature.

Evro isn't built on intuition about what "good communication" looks like. Every feature is grounded in peer-reviewed research.

📄

Cognitive Load Theory (Sweller)

How Evro handles the monitoring burden so your brain can focus on the conversation.

📄

Deliberate Practice (Ericsson)

How every meeting can become a real-context improvement session.

📄

Gottman Interaction

How Evro tracks whether the relationship is on track or not based on established research.

What Evro is, and isn't.

Some tools track employees. Some monitor team performance. Evro is not one of those tools. Here's the full picture.

Communication coaching

Evro helps you understand how you communicate and improve over time privately.

Meeting memory

Evro captures what was said and what was agreed so you're never reconstructing from memory.

Relationship intelligence

Evro builds a living understanding of every person you work with for your benefit, not your organisation's.

Not employee surveillance

Evro does not report your performance to your manager, HR, or employer.

Not a team monitoring tool

There are no team dashboards, no leaderboards, no manager-visible scores.

Not a data scraper

Evro is not tool to harvest data and sell it to others. It's a communication tool built on trust, not breaking your trust.

The people behind Evro.

Evro is being built by experts who want to solve the problems of communication.

Jay Spence

Jay Spence

Co-Founder & CEO

PhD and Masters in Clinical Psychology. Two decades of experience as a therapist. A decade working in corporate settings at C-level. Bringing the scientific frameworks and clinical experience around effective communication.

Elaine Lee

Elaine Lee

Co-Founder & CEO

Three decades of corporate experience across startups to Fortune 500 orgs from junior level to head of division. Two decades of experience in product and engineering. Brings the translation between science and software with safety in mind.

Privacy questions people ask most.

I. Data Training & Isolation

No. We default to a zero-training policy. Your inputs and outputs are never used to train our base models or those of our providers.
Impossible. Because your data is never used for training, it cannot "leak" into the responses generated for other users.
We use logical multi-tenant isolation and role-based access controls, ensuring your data can only be accessed by you.
We don’t use your data to improve models of any kind. No such option is offered.

II. Third-Party Sharing

Never. Our business model is based on subscriptions, not selling user data or behavioral insights.
We only use audited partners (like OpenAI or Anthropic) who are contractually bound to the same "zero-training" and privacy standards we provide you.
Yes. Our full list of subprocessors is maintained in our Trust Center and updated in real-time.
No. Human review only occurs if you explicitly flag a specific response for a support ticket.

III. Agentic Autonomy & Control

No. There are no “Agents” live in Evro at this stage. If built in future, you will be informed. Our plan is for every agentic action to have a "Human-in-the-Loop" by default. You would approve any action that interacts with outside systems.

IV. Data Deletion & Persistence

Yes. Deletion triggers a purge from our live databases immediately and from encrypted backups within 30 days.
Yes. When you delete data, we programmatically remove the associated vector embeddings to ensure the AI "forgets" that context.
We retain data for 30 days post-cancellation to allow for recovery, after which all customer data is permanently wiped.

VI. Anonymity & Profiling

No. We do not track your emotional state or interests for marketing; we only store history to provide you with a persistent product experience.

VII. Security & Attack Protection

We use a dual-layer firewall: a "Guardian Model" inspects every prompt for malicious instructions before it reaches the main AI.
Yes. All data is encrypted using AES-256 at rest and via TLS 1.3 in transit.