How we build, deploy, and govern Maeve — our AI guidance assistant — responsibly. Including our approach to automated decisions, human oversight, and data privacy.
Our position in plain English: Maeve exists to help — not to replace human judgement, not to make decisions, and not to handle what should be handled by a person. Bereavement is one of the most sensitive moments in a person’s life. We take that seriously in every design decision we make.
besideU's platform does not make, inform, or record any employment decision, performance assessment, absence determination, or HR outcome through automated means.
All decisions relating to any employee remain the sole responsibility of the employer. This is not a limitation of the product — it is a deliberate design principle embedded in Maeve’s architecture from day one.
Maeve does not perform automated decision-making or profiling as defined under UK GDPR Article 22. Individuals are not subject to decisions based solely on automated processing that produce legal or similarly significant effects.
Maeve is besideU’s AI guidance assistant. She is designed to help managers, HR teams, and employees navigate bereavement at work — providing practical, policy-specific guidance at the moment it’s needed most.
Maeve is trained on UK bereavement best practice, the Employment Rights Act 2025, ACAS guidance, and your organisation’s specific bereavement policy. She provides contextual, practical guidance — not generic advice.
Important: besideU is not a clinical service. If anyone using the platform is experiencing a mental health crisis, Maeve will signpost to appropriate professional support — including the Samaritans (116 123), Mind (0300 123 3393), and your organisation’s EAP. She will not attempt to provide crisis counselling.
Every recommendation Maeve makes is advisory. Humans make the decisions. Maeve supports that process — she does not replace it.
Bereavement data is treated as sensitive regardless of its technical classification. We design every interaction with that in mind.
Users always know they are interacting with an AI. Maeve never presents herself as a human. Her nature, capabilities, and limitations are clearly communicated.
Maeve only accesses and uses data that is necessary to provide relevant guidance. She does not profile users beyond what is needed for the task at hand.
Data from Maeve interactions is never used to train external AI models. Your data stays within your organisation’s environment.
Maeve is designed to recognise when a situation exceeds what AI guidance can appropriately address and to escalate to HR or professional support accordingly.
besideU applies responsible AI principles designed to keep Maeve safe, transparent, and appropriately bounded in emotionally sensitive workplace contexts. These commitments are built into the product architecture — not added retrospectively.
Human oversight — all AI-generated guidance operates within defined escalation thresholds that prioritise human intervention. Maeve will not attempt to resolve situations that require a human professional.
Bounded AI behaviour — Maeve’s scope is explicitly constrained. She operates within a defined set of tasks and will not speculate, improvise, or attempt to perform actions outside her design parameters.
Escalation-first safety design — where Maeve is uncertain, she escalates. Conservative thresholds ensure that ambiguous situations default to human involvement rather than AI-led resolution.
Privacy by design — data minimisation is a core architectural principle. Personal data is not retained beyond what is necessary. Client data is never used to train or fine-tune Maeve.
Output monitoring — Maeve’s outputs are regularly reviewed against documented behavioural and safety criteria, including monitoring for quality, accuracy, and potential bias.
Annual policy review — this governance framework is reviewed annually and promptly in response to material changes in regulation, platform capabilities, or the nature of how Maeve is used.
Our governance approach is informed by the following standards. We do not claim formal certification against all of these, but they actively shape how we design, monitor, and govern Maeve.
We are working towards ISO/IEC 42001 certification as a formal AI management systems standard. We will update this page as our certification status progresses.
Maeve uses large language model (LLM) technology to generate contextual guidance based on the information provided in each interaction. She is configured with:
By default, Maeve conversation content is not retained beyond the session. Where retention is enabled — for example to support HR oversight of manager interactions — this requires explicit agreement from your organisation and is governed by your Data Processing Agreement.
Conversation data is never used to train third-party AI models. Any use of data to improve Maeve’s responses is done only within your organisation’s data environment and only with your explicit consent.
Like all AI systems, Maeve can make mistakes. She may occasionally provide guidance that is incomplete, out of date, or not appropriate for a specific situation. This is why:
In addition to guiding managers and HR teams, Maeve serves a second and distinct role — as a compassionate AI grief companion for bereaved employees. This is a deliberate and carefully governed part of the besideU platform, and the first of its kind in a UK workplace context.
Maeve is available to employees throughout their bereavement journey — from the moment of loss through leave, return to work, and beyond. She can help employees understand their entitlements, find the right words for difficult conversations, and access appropriate resources. She is available 24 hours a day, including outside working hours when professional support may not be accessible.
Important distinction: Maeve in her companion role is not a counsellor, therapist, or crisis service. She is a knowledgeable, compassionate presence that listens, informs, and signposts. Every interaction is governed by the same strict ethical guidelines that apply across the full platform.
Maeve operates a traffic light system to govern how she responds to employee interactions, particularly when conversations touch on emotional distress or wellbeing:
| Level | What it means | What Maeve does |
|---|---|---|
| 🟩 Green | Normal grief journey — employee is coping, seeking information or practical guidance | Provides guidance, answers questions, signposts resources, maintains warm companionate tone |
| 🟡 Amber | Signs of elevated distress — language suggesting the employee is struggling significantly | Acknowledges distress, gently encourages professional support (EAP, Cruse, Mind), flags the interaction to HR in an anonymised summary |
| 🔴 Red | Indicators of crisis — language suggesting risk to self or others, acute mental health distress | Does not attempt to manage the situation. Immediately signposts to crisis services (Samaritans 116 123), notifies HR directly, and ends the AI-led interaction |
The traffic light system is reviewed regularly by the besideU team and governed by our clinical advisory framework. Thresholds are deliberately conservative — Maeve escalates early rather than late.
Not everyone on the platform sees the same data. Access is strictly role-based:
Maeve never contacts employees unsolicited. She responds when spoken to — she does not initiate contact or send unprompted messages.
besideU is committed to ensuring Maeve treats all users equitably regardless of the nature of their bereavement, their relationship to the deceased, their cultural background, or their personal circumstances.
We recognise that bereavement is experienced differently across cultures, religions, and personal circumstances. Maeve is designed to be sensitive to this — she does not assume a single model of grief or a single definition of family.
We regularly review Maeve’s outputs to identify and address any patterns of bias or inconsistency. If you identify a response from Maeve that feels biased, inappropriate, or culturally insensitive, please report it to us at [email protected].
This AI Ethics Policy is reviewed at least annually and whenever we make significant changes to how Maeve works. We are committed to keeping it current as AI technology and regulation evolve.
besideU monitors developments in UK AI regulation — including the government’s AI Opportunities Action Plan and any future AI legislation — and will update our practices accordingly.
We welcome questions, challenges, and feedback on how we use AI. Responsible AI is not a destination — it is an ongoing commitment.
If you have questions about this policy, concerns about how Maeve has behaved, or want to understand more about how AI is used in besideU, please get in touch.
Email: [email protected]
Subject line: AI Ethics Enquiry
besideU Ltd · Registered in England and Wales
For data protection queries specifically, please see our Privacy Policy.