Responsible AI

LEVRA uses AI powered simulations to help people practise conversations in a safe, judgement free space.
We wanted to provide a simple summary of how it works and what data is used.

Video call interface showing a man with glasses and white hoodie with realtime skills feedback panels.

A place for practice

Our AI avatars offer a space to practise and rehearse conversations. They are not an authority. They help learners build Human Skills such as communication, judgement and empathy through realistic scenarios. They supplement rather than replace professional advice or company policy.Once a learner completes a simulation, AI generated scores and feedback are provided so people can learn from, develop and improve. LEVRA never uses AI outputs or results for hiring, promotion, performance ratings or any other employment decision.

You always know you're talking to AI

Every simulation is clearly presented as an AI powered exercise. Each avatar follows a defined character, tone and set of boundaries. It cannot go off topic or present itself as anything other than a training tool.

How AI is involved
^{Our simulations run on OpenAI's large language models. We design the scenarios, the prompts, and the guardrails. In practice, this means:}
‍

Personal data is never used to train OpenAI's or any other third party's AI models.

The AI only receives the text of a learner's spoken input. It never receives a learner's name, email or profile details.

Conversation transcripts are kept in pseudonymised form only, linked to a system ID rather than a name, and only for the length of our contract with our AI provider.

Transcripts can be deleted on request and are used only to run the system, check quality and support moderation.

Moderation and human oversight

Every completed simulation is automatically screened to help identify inappropriate or unsafe outputs such as harassment, hate, self harm, sexual content and violence. If something is flagged, authorised LEVRA team members will be made aware to review the automated moderated checks.

Keeping your data safe

Personal data is stored and processed in UK based Microsoft Azure infrastructure, within an ISO 27001 aligned security framework. Pseudonymised AI conversation data is processed by our simulation provider in EU based infrastructure. Any transfer outside the UK or EU is protected by appropriate legal safeguards.Our security measures include multi factor authentication, encryption in transit and at rest, segregated environments and continuous monitoring. We have completed a Data Protection Impact Assessment covering our platform and AI simulations, and review it regularly.

Responsible AI

LEVRA uses AI powered simulations to help people practise conversations in a safe, judgement free space.
We wanted to provide a simple summary of how it works and what data is used.

A place for practice

You always know you're talking to AI

How AI is involved
^{Our simulations run on OpenAI's large language models. We design the scenarios, the prompts, and the guardrails. In practice, this means:}
‍

Personal data is never used to train OpenAI's or any other third party's AI models.

The AI only receives the text of a learner's spoken input. It never receives a learner's name, email or profile details.

Conversation transcripts are kept in pseudonymised form only, linked to a system ID rather than a name, and only for the length of our contract with our AI provider.

Transcripts can be deleted on request and are used only to run the system, check quality and support moderation.

Moderation and human oversight

Keeping your data safe

Your control

Learner data is pseudonymised,
so it cannot be connected back to an identifiable individual without
a secure key.

Your organisation can request deletion of any information at any point, subject to the retention limits below.

Clear retention limits apply across every service we use.

We support access requests, corrections and deletions in line with GDPR timelines.

Please fill in your details to download the report.

Responsible AI

LEVRA uses AI powered simulations to help people practise conversations in a safe, judgement free space. We wanted to provide a simple summary of how it works and what data is used.

A place for practice

You always know you're talking to AI

How AI is involvedOur simulations run on OpenAI's large language models. We design the scenarios, the prompts, and the guardrails. In practice, this means:‍

Personal data is never used to train OpenAI's or any other third party's AI models.

The AI only receives the text of a learner's spoken input. It never receives a learner's name, email or profile details.

Conversation transcripts are kept in pseudonymised form only, linked to a system ID rather than a name, and only for the length of our contract with our AI provider.

Transcripts can be deleted on request and are used only to run the system, check quality and support moderation.

Moderation and human oversight

Keeping your data safe

Your control

Learner data is pseudonymised, so it cannot be connected back to an identifiable individual without a secure key.

Your organisation can request deletion of any information at any point, subject to the retention limits below.

Clear retention limits apply across every service we use.

We support access requests, corrections and deletions in line with GDPR timelines.

Please fill in the form and we will get in touch shortly!

LEVRA uses AI powered simulations to help people practise conversations in a safe, judgement free space.
We wanted to provide a simple summary of how it works and what data is used.

How AI is involved
^{Our simulations run on OpenAI's large language models. We design the scenarios, the prompts, and the guardrails. In practice, this means:}
‍

Learner data is pseudonymised,
so it cannot be connected back to an identifiable individual without
a secure key.