Serene and AI, in plain words
Version v1.1 · Last updated 18 October 2026
What Serene is
An AI chat that helps you think through a hard moment, a letter or a worry, and points you to real help. It's built with Claude Opus 5.5, an AI model made by Anthropic, running on Amazon Bedrock, part of Amazon Web Services, in London. Serene is not a person. It can be wrong. It isn't medical, legal or financial advice, and it's not for emergencies. Serene is a wellbeing and signposting tool. It does not assess, diagnose, monitor or treat any condition.
Right now
Serene is live. It is an AI, not a person. For adults 18+ only. Not a medical professional.
Who can use it
Adults 18 and over, signed in, who ticked the Serene box (you can change it in Settings). The Serene chat needs a Supporter plan. The AI checks (Check a message, Check a payment, Scan & remind and advice letter drafts) are free, with a daily or monthly limit. Accounts start at 13; Serene and the Supporter plan are for adults 18 and over. Your age is what you tell us; we do not check it. Never young person accounts for Serene, never child-profile phones.
What it costs
Every safety tool in SereneHelp is free for everyone, always. Serene and the Supporter plan are for adults 18+ only.
What we send
Your new message (and a picture if you add one), up to your last 6 messages in that chat (shortened), and our safety instructions. Earlier messages stay only on your screen while the chat is open; we never store them. Never your name, email, contacts, Crisis Card or anything else on your phone.
Where it goes
Your words pass through our app platform, Base44, on the way. Base44 stores account data in the UK, and runs our code and email from US servers. They then go to Amazon Web Services in London (eu-west-2), where the model replies. Base44 says that on plans below Enterprise, which includes ours, data submitted through an app can be used to train AI models. We are asking Base44 for a written agreement that SereneHelp data is never used this way. Until we have it, this risk is real, and we tell you so.
What's kept
We don't keep your chats; when you close a chat, it's gone. Amazon Bedrock doesn't store your messages or replies for this model and doesn't use them to train AI models. Anthropic never receives them. One exception: if Amazon's automatic systems flag what looks like child sexual abuse material, Amazon may keep it and report it to the authorities.
What we record
How many messages you've used this month and their estimated cost, plus monthly totals for all of SereneHelp with no names. Never your words.
How we keep it safe
Your words are checked on your phone for signs you're struggling. That changes only the order and tone of Serene's answer; every answer offers 999, Reach a person and My plan. Serene never names or describes methods of harm. We follow NICE guideline NG225 and Samaritans' guidance on safe content; they haven't reviewed or endorsed SereneHelp. If the model or our instructions change, Serene switches off until its safety tests pass again.
Other AI in SereneHelp
Check a message, Check a payment, Scan & remind and advice letter drafts use the same AI model, on Amazon Web Services in London. Each sends only what you paste, type or photograph, once, and we keep nothing of it. No other AI is used.
Something wrong?
Email safety@
Crisis and helpline information may be translated automatically. If you are in danger or need urgent help, contact your local emergency number straight away. To report a mistake in safety or helpline information, email safety@serenehelp.org.
This is a self-help tool, not therapy or a diagnosis.