Mila is an AI, not a human and not a clinician. She is designed to be warm, evidence-based, and gentle. She is not a substitute for your paediatrician, and she cannot diagnose or treat any medical condition. See our Medical Disclaimer for the full boundary.
What Mila is
Mila is the name we’ve given to the AI companion built into GentleMonths. She is the “voice” of the brand — the tone in which we speak with you, whether you are reading your personalised report, receiving an email, or chatting inside the GentleMonths Club.
Under the hood, Mila is built on large language models (LLMs) developed by Anthropic, PBC — specifically the Claude family of models (Haiku, Sonnet, and Opus). We selected Anthropic because of their documented commitment to AI safety research and because they do not train their foundation models on customer data by default.
What Mila is not
- Mila is not a person. She has a name and a personality, but she is a piece of software.
- Mila is not a doctor, nurse, or lactation consultant. She holds no licences and cannot practice medicine.
- Mila is not a therapist. For anxiety, depression, or any mental-health concern, please reach out to a qualified professional.
- Mila is not an emergency service. In a medical emergency, call 911 (US), 999 (UK), 112 (EU), or your local equivalent immediately.
How Mila works
1. Personalised context
When you complete a quiz, we build a small structured context for Mila: your first name, your baby’s age in weeks, your parenting archetype, your main concerns, and your specific quiz answers. No more than that.
2. Guided prompting
We then ask Claude, in a carefully engineered prompt, to produce your personalised content — while respecting a detailed voice and safety brief that we have tuned over many iterations. Claude generates text; we never hand you the raw model output without safety checks.
3. Structured output & validation
Responses are returned in a structured format (JSON with known fields) that we validate against a schema before rendering anything to you. This reduces the chance of malformed or off-topic output reaching your inbox.
4. Human review on the risky parts
High-stakes content — such as our Your Money or Your Life (YMYL) articles on sleep, feeding, and development — is reviewed and edited by a human editor before it is published. Short personalised report sections are AI-authored; the underlying templates, archetype logic, and safety rules are designed and maintained by us.
What Mila can help with
- Identifying which developmental phase your baby is in and what it typically looks like.
- Explaining the reasoning behind common gentle-parenting approaches.
- Suggesting a starting point for sleep, feeding, and emotion-regulation strategies — grounded in your baby’s age and your situation.
- Helping you draft partner/family conversations (e.g. explaining your approach to a grandparent).
- Cross-referencing information against well-known authorities (AAP, WHO, NHS) so you know where a claim comes from.
What Mila will not do
- Diagnose a medical condition or suggest medication/dosage.
- Give advice that overrides your paediatrician’s instructions.
- Comment on a specific symptom as if it were a clinical evaluation.
- Shame any parenting choice, including ones we would make differently.
- Claim to be human, to be a doctor, or to be licensed in any profession.
Your data and Mila
Here is exactly what happens with the information you share:
- Your inputs are sent from our servers to Anthropic’s API over encrypted channels.
- Anthropic processes the request as our data processor and returns the completion.
- Anthropic retains request data for a short abuse-prevention period (typically 30 days) and then deletes it.
- Anthropic does not use this data to train future foundation models.
- We store the resulting report in your account on our Neon PostgreSQL database (EU-hosted) so you can come back to it.
Full details, including your rights under GDPR, are in our Privacy Policy.
AI transparency and compliance
We comply with applicable transparency rules for AI-generated content, including the U.S. Federal Trade Commission’s guidance on AI disclosures and Article 50 of the EU AI Act. That means you will always see an “AI-generated by Mila” indication on AI-authored content, and you can always ask us to confirm whether a specific piece of content was generated by AI, edited by a human, or both.
Feedback and errors
If Mila ever produces something that feels wrong, inappropriate, or potentially harmful, please tell us. We investigate every report and use it to improve the guardrails. Our safety address is [email protected].
You can also request that we retrain, adjust, or delete a specific response. We take every request seriously.