Slow essays on AI & health, research, strategy, ethics, tools, and the future. Calm, evidence-based, premium. No newsletter spam.
Featured
Forbes just rounded up the most accurate sleep trackers of 2026 — rings, watches, brain-sensing headbands, sleep mats, even your phone. Impressive sensors, every one. But accuracy is not understanding, and a better score is not a better night. Here is the part the device makers keep getting backwards.
All essays
131 essays
A prompt-to-3D tool for creating small, interactive visuals that make your wellness data and protocols less abstract.
The build a small clinic actually ran, in order — not a tool per leak, but one seat set up like a role. Starting from a cold inbox, working through the handbook, the read-only connector and the rooms, and implemented as a closed gap between demand and care with a human still holding the pen.
The legal automation platform is built for elite law firms, but I found a narrow, high-value use case for wellness practitioners.
The build, not the pitch — the exact sequence a solo practitioner ran to induct a worker as an associate. Starting from an empty second chair, working through the draft-and-sign-off loop that keeps care human, and implemented as recovered evenings and follow-ups that finally get sent.
You don't need another platform. You need a better process for the AI tools you already have.
This is the build, not the brochure — a real sequence you can copy. Starting with one unread export, working through the exact moves that make a worker read it honestly, and implemented as a standing Sunday reading that hands you a shortlist and leaves the decision where it belongs: with you.
A small wellness clinic was losing clients in the space between the enquiry and the booking — not for lack of care, but for lack of hands. Here is what changed when they stopped buying a tool per leak and set up one seat, inducted like a role, with a human holding the pen.
Mem is an AI-first notebook that automatically organizes your health data, making symptom tracking and pattern recognition truly useful.
A one-person practice was drowning at the edges — the intake skimmed at 9pm, the follow-up that never got sent, the prep done in the car park. Here is what changed when the practitioner inducted a worker as an associate, kept a human at the sign-off gate, and got the evenings back.
Claude excels at calmly reasoning through complex health information, from lab results and research papers to your own symptom journals.
One person, four years of scattered health data, and a nagging sense that the answer was already in there somewhere. Here is what changed when she stopped scrolling through it and handed the whole pile to a worker that could hold it all at once — and why the useful part was still the decision she made at the end.
Skip the gurus and health apps. Learn to build your own longevity strategy using the AI tools you already have.
One is a free public library of nearly every medical paper ever published. The other is an AI answer engine that reads those papers for you and writes back a referenced conclusion. They are not competitors so much as two ends of the same job. Here’s what each one actually does, whether OpenEvidence is free, how accurate it really is, and how to use both inside the Research Layer without handing over your judgement.
A clinic does not need ten AI tools. It needs one seat, set up like a role — in about an hour. Write the operations handbook, wire one read-only connector, build a room per workflow, schedule the drudge, and put a human at the sign-off gate. Here is the first-hour setup for a wellness business.
The associate you could never afford does not arrive trained — you induct it. One focused hour: write the practice induction, upload a real intake, build a room per client, and wire the sign-off gate before anything reaches a soul. Here is the first-hour setup for a one-person practice.
This open-weights reasoning model offers a compelling, self-hostable alternative for building your personal AI health stack.
You do not need to learn the tool. You need to set it up once. Sixty minutes, four moves — write the brief, pour in your data, build one room, set one standing job — and the chat box turns into something that quietly reads your own body for you. Here is the first hour, step by step.
Learn a simple method to turn your existing AI tools into a powerful, private health ledger.
This tool lets you design and run automated, multi-step health research and data analysis workflows that operate continuously in the background.
A clinic does not need ten AI tools. It needs one seat, set up properly. Induct it once with an operations handbook, connect it to your inbox and calendar, schedule the drudge work, and put a human at the sign-off gate. Here is how the features map to the roles your business keeps dropping.
How to use the AI tools you already have to get more from your clinical care.
The model is not the hire — the features are. Memory holds the whole client, uploads read the real file, projects give each client a room, and connectors reach into the tools you already run. Here is which switches a one-person practice should turn on, and where the human stays in the loop.
This background agent automates the tedious, multi-step research and drafting work that underpins any serious wellness plan.
Two AI research tools keep coming up when people want to check what the evidence actually says about a supplement, a symptom, or a protocol. One is built for clinicians and reads like a specialist. One is built to summarise the literature and reads like a very fast librarian. Here’s how they differ, where each one earns its place, and how to stack both inside the Research Layer without handing over your judgement.
Everyone talks about the model. Almost nobody uses the features. Fable 5’s real leverage isn’t the answers — it’s memory, uploads, projects and scheduled actions. Here is a plain-English tour of the switches that turn a chat window into something that quietly reads your own body for you.
Vercel’s v0 is an AI code generation tool that lets technical users prototype their own health and wellness interfaces.
Clinics and studios do not fail from a lack of demand. They stall on the admin between the demand and the care. A large-context model is the first role you can staff without payroll — if you give it a desk, a job description, and a human who signs off.
A step-by-step guide to finding evidence, not just hype. Learn the prompts that separate signal from noise in the crowded world of dietary supplements.
The company that sells the reasoning tool to pharma now wants to own the molecule too. Anthropic launched Claude Science for drugmakers and, in the same breath, said it will start discovering drugs of its own. It’s a clean example of the thing we keep flagging: AI money entering health almost never has a single motive. Here’s what vertical integration of intelligence actually means for you.
A model that can hold a client’s whole history at once is not a smarter search box — it is the associate you could never afford. Here is how to delegate the draft, gate the decision, and end up with more time for the part of the work only you can do.
Midjourney can translate abstract health data and goals into a tangible visual reality, adding a powerful new layer to your AI health stack.
A new class of European health copilot has found the sharpest pain in the whole category and built a beautiful app around it: three apps, one blood report, zero answers. The pitch is right. The finish line is wrong. Here's what the download-an-app model gets right — and the dependency it quietly rebuilds.
The frontier models can now hold a whole day of work in their head at once. That upgrade quietly changes what they are for. Fable 5 is not a smarter thing to chat with — it is the first one you can hand a real job to. Here is what that means when the job is understanding your own body.
Fyxer learns your writing style to draft email replies, triage your inbox, and summarise long threads, but does it work for wellness workflows?
Go beyond symptom trackers and build your own private health AI with your medical history and lab reports.
OpenAI Operator can automate the tedious web-based tasks of health management, from researching supplements to filling forms.
This networked note-taking tool acts as a private, searchable memory for your health journey, turning scattered data into clear insights.
Florida State watched C-grade students turn their results around in weeks with it. The reason isn’t the podcast feature everyone talks about — it’s that NotebookLM only answers from the sources you give it, and will tell you when it doesn’t know. Here is the full deep dive: what it actually does, why grounding is the whole point, and how to use it for your own health, not just a final exam.
Stripe, Anthropic and the OpenAI Foundation are putting half a billion dollars behind a nonprofit to end the common cold. It’s real, it’s not hype, and it tells you exactly how AI is arriving in health: from the top, through capital and protein design and the air in your office — not through another app on your phone. Which is precisely why the part you control matters more, not less.
“AI for stress relief” is sold as a feeling — open an app, breathe, feel calmer. The honest version is narrower and far more useful: AI is excellent at a few specific stress jobs and useless, sometimes harmful, at the rest. Knowing the difference is the whole skill. Here is the clean split — what AI can do for stress, what it can’t, and how to use the can without sliding into the can’t.
The decision to lean on an AI for something tender almost never gets made on purpose. It happens at midnight, mid-spiral, with a tool that is fluent, patient and infinitely agreeable — three qualities that feel like care and aren’t. This is the pre-flight: seven questions to run before you trust an AI with your mental health, so the choice is one you made rather than one that happened to you. None of them takes longer than the spiral already will.
“Which AI is best for anxiety?” is the question everyone types and almost no one answers honestly, because the honest answer is a question back: best at what? There is no anxiety model. There is a research tool, a structuring tool and a drafting tool — and on an anxious night you reach for whichever one is open and hope it does all three. It won’t. Here is how to route the actual job to the actual tool, why the brand barely matters, and the one job no AI should ever take.
Perplexity delivers sourced, verifiable answers, making it an essential first layer for any evidence-based wellness query.
This text-to-video model is less about cinematic storytelling and more about creating powerful, personalized health visualizations.
A practical guide to analyzing your personal health metrics using the large language models you already have.
This HIPAA-compliant AI scribe automates the most tedious part of a practitioner’s day: turning patient conversations into structured clinical notes.
How to use X-dot-com’s real-time AI to track—and gut-check—trending health narratives without the noise.
Most journaling asks you to narrate your feelings. Your nervous system doesn't speak in narratives — it speaks in states: settled, mobilised, shut down. When you log the state instead of the story, two things happen. The entries get short enough to actually keep, and the pattern becomes legible to you and to an AI you can ask about it. Here's how to run a state-based ledger that tracks your baseline instead of your mood — and why it reads your data better than any app built to track your output.
The interesting AI story this month isn't a new chatbot. It's that the office suite you already own quietly learned to brief your day, prep your meetings, and block your recovery time — no extra app, no new login. That's the same shift we keep pointing at: AI moves into the tools you already use. Here's how to put that shift to work for your health instead of your inbox — and the governance questions nobody in the productivity blogs is asking.
Before you add that bio-wearable to your cart, here’s a radical idea: the most powerful health tool is already on your phone. And it’s free.
This wearable isn't just for tracking sleep; it’s a tool for building a more intelligent, data-informed wellness practice.
Your phone isn’t just changing your attention span — it’s changing the physiology underneath it. The same nervous system that decides whether you can finish a paragraph also decides how you sleep, digest, and recover. So here’s the uncomfortable reframe: attention is not a productivity problem, it’s a vital sign. And the tools rebuilding it are the same ones quietly capturing it. Here’s how to read your own attention the way you’d read your HRV — and use AI to do it without handing the data back to the machine that took it.
Generative image models can now read, annotate, and even fabricate medical-looking imagery with unsettling fluency. That cuts two ways: a genuine assistive leap for people learning to understand their own scans, and a new frontier for confident, beautiful, completely wrong pictures of your insides. Here’s the honest line between what AI imaging can do for your health stack and where it quietly becomes a liability — and how to use it as a reading partner without mistaking a render for a diagnosis.
How to use Notion AI to connect your health records, research, and personal protocols into one intelligent system.
When the people building the most capable AI on the planet say they expect to help cure most disease within a decade, two reactions are equally useless: breathless belief and reflexive cynicism. The interesting move is to take the claim seriously and literally at once — to ask what would actually have to be true, what it changes for you now, and what it conveniently leaves out. Here’s a calm read on the most ambitious promise in health AI, and what a sensible person does with it this year.
Suno generates custom, royalty-free audio from text, creating new possibilities for personalized wellness soundtracks.
Your heart rate, your sleep stages, your cycle, your glucose, your voice, soon your gait and your face — the most intimate dataset that has ever existed is being assembled, and almost none of it is held by you. AI makes that data dramatically more valuable and dramatically easier to read, which raises a question the terms-of-service skip past: who actually owns the record of your own body? Here’s a clear-eyed look at biodata ownership in the AI era — and the practical, unglamorous way to take your copy back.
Build a personal evidence library, understand complex topics, and partner with your clinician—without needing a PhD to read the primary literature.
The design tool you already know is now a surprisingly fast way to visualize your health data and create polished, motivating protocols.
A browser-based coding environment with a built-in AI agent that helps you build custom scripts and tools for your health data.
“Is AI therapy safe?” is the wrong question with the right instinct behind it. There is no single answer because there is no single thing called AI therapy — there is a branded bot, a general chat tool, and a human practitioner with software, and they fail in completely different ways. Here is the honest version: where an AI conversation is reasonably safe, where it quietly is not, which AI is actually best for the job, and the four questions to ask before you trust any of them.
“Which AI is best for health?” is the wrong question, because it assumes one model wins everything. It doesn’t. ChatGPT, Claude, Gemini and Perplexity each have a job they’re genuinely better at — and the trick isn’t picking a favourite, it’s knowing which one to reach for at which layer of your health stack. Here is the honest division of labour, mapped to the 3-Layer Method, so you stop switching tabs at random.
This popular sports watch is a powerful data source for your AI health stack, if you know how to use it.
Its massive context window and document analysis capabilities make it a uniquely powerful tool for patient-led literature reviews. Here's how to use it.
This AI search engine, designed for clinicians, provides cited, evidence-based answers for deep wellness research.
A post doing the rounds claims Claude has a hidden ‘ADHD Executive Function Mode’ that lets you finish a week of work in four hours. It doesn’t. There is no secret mode. But the seven prompts underneath the hype are genuinely good — because they do the one thing a scattered brain can’t do for itself: hold the structure outside your head. Here’s how to set it up properly.
Every few months the names change and the launch posts insist this one is the smartest model ever made. For running your own health data, that’s the wrong question. There aren’t seven models to memorise — there’s one ladder, and a simple rule for which rung fits which job. Here’s the calm version: what Haiku, Sonnet and Opus actually are, and which one to reach for when you’re reading your own labs, sleep and notes.
Forget presets and dashboards. Learn to analyze your raw sleep data from wearables to find patterns that actually matter.
Pasting your wearable data into a large language model is a popular experiment. Here’s what actually works, what doesn't, and how to get useful insights instead of generic replies.
A continuous glucose monitor is not a score to be optimized, but a powerful tool for learning the language of your own metabolism.
The clever prompt was a 2024 skill, and it's being automated away. The durable skill in 2026 is briefing an agent, directing it across an entire system, and curating what comes back. Here is what that shift means for your health, your work, and why taste — not prompting — is the thing worth keeping.
The accounts winning the wellness feed aren't more correct than the practitioners losing it — they're more available and more certain. You can't beat entertainment dressed as knowledge by being louder. You beat it by being recognisable, consistent, and believed. Here is the branding method for time-poor practitioners, and the part of it AI now does for you.
We tested the new AI features for reviewing complex health agreements, so you know what you’re actually signing.
Synthesia creates studio-quality AI avatar videos from text, offering a powerful tool for practitioner training and client communication.
Canva's AI-powered design tools make it surprisingly simple to create clear, actionable health and wellness materials.
It’s not a doctor, but it is a powerful reasoning partner for your wellness journey.
With Apple Intelligence, the simplest note-taking app on your phone becomes a surprisingly capable tool for wellness tracking and synthesis.
A step-by-step method for interpreting your lab results with AI, plus the critical safety checks you can't skip.
Elicit acts as an AI research assistant, helping you conduct systematic literature reviews across millions of academic papers.
Otter AI transforms spoken audio into a structured, searchable text archive for your personal and professional wellness work.
The speed-focused email client introduces AI drafting and triage, but how does it perform in a serious wellness workflow?
Stop downloading apps. Learn to build a private, powerful, and personal health ledger with the AI you already use, no subscription required.
AI is not the enemy. Surrendered agency is. A ten-point reading of Pope Leo's Magnifica Humanitas, translated out of theology and into the only thing that matters for your health: who is allowed to read your body, and who decides what happens next.
Decagon builds AI agents to automate the repetitive client questions that consume a practitioner's day, freeing you for high-value clinical work.
Your clients try every protocol from the feed. They feel something. They report something. But you can't follow twelve people around all week asking how their magnesium is going — your liability insurance doesn't cover stalking. And no app rolls their messages, wearables, and food photos into one pattern you can act on. This is the practitioner job AI literacy actually does.
You try every longevity hack from the feed. You feel something. You see something. But nothing tells you which intervention is actually working — and you don't have a health team following you around with a clipboard. (Unless you do, in which case — respect.) Here's the part the algorithm can't do for you.
This AI voice tool can read your research, notes, and protocols back to you — in your own voice or a custom one.
Most AI health coaches are just chatbot wrappers you pay a subscription for. We'll show you how to build a more powerful, private, and personalized system for free using the AI tools you already have.
Most practitioners answer client messages from four places — inbox, DMs, WhatsApp, the booking platform — and lose ninety minutes a day to re-typing the same five paragraphs. Wire a reasoning chat into the messenger your clients already use, with a standing instructions block written for your practice, and the average touch drops from six minutes to ninety seconds. The shape of the setup, what it does, and where it stops.
Most people open a separate app to talk to an AI, then a second app to log it, then forget the third app where the notes were supposed to go. Wire a reasoning chat tool into the messenger you already use all day and the whole stack collapses to one thread. Captures land in your ledger. Questions trigger background research. Follow-ups send themselves. No new app, no new dashboard — and you stop losing the thought between opening apps.
Eleven SaaS subscriptions, four launch courses, two years of "experiments". Roughly £14,800 in a year, one underperforming launch to show for it. Then she replaced almost all of it with a single 17-step content cycle. The shape of the system, what stayed, what got cut, and why the cheapest line item in a practice is the one that replaces eleven others.
Use Google-powered AI for live, cited answers to your complex wellness questions, directly from search results.
Stop downloading health apps. Start building a system that learns you. This guide shows you how to use the AI tools you already have to manage your health data.
This professional-grade AI video suite is surprisingly useful for creating tangible, visual representations of your wellness data.
Krea's real-time image generation offers a surprisingly fluid way to visualize health data, client concepts, and personal wellness protocols.
A calm look at building bespoke wellness tools with natural language, moving beyond spreadsheets and generic apps.
Moving beyond chatbots to create a branded conversational AI that can handle client communications for clinics and wellness brands.
This is not another app. It's a framework for using the AI you already have to understand your health—intelligently and safely.
Zapier’s AI agents move beyond simple triggers, letting you build conversational workflows that automate health data analysis and client management.
Generate polished, multilingual presenter videos from a script, scaling your ability to explain complex health topics without endless re-recording.
Devin, the autonomous AI engineer, promises to build applications from a single prompt. I tested it on a real wellness workflow.
Mistral's Le Chat offers a fast, private, and EU-hosted AI chat for your daily wellness writing and research.
NotebookLM turns your health documents into structured audio, summaries, and insights, grounded only in the data you provide.
People are quietly using ChatGPT as a 3 a.m. listener, a CBT scratch-pad, and a place to draft the email they cannot send. Some of that is fine. Some of it is not. A short, honest framework for what general-purpose chat is genuinely good at, where it is dangerous, and the four-line script we use to keep a chat thread on the right side of that line.
It provides a structured, longitudinal record of your biology, forming a critical data layer in a personal AI health stack.
Slides, explainer videos, podcasts, and infographics used to be four tools and four afternoons. In 2026, one document becomes all four — in one tab, before lunch. Here's what changed, what's actually worth keeping, and the three tools we'd build the year around.
A solo practitioner replaced a stalled 12-week launch with three pages, one free 10-day email course, and a three-pillar social rhythm — and started filling discovery calls in a fortnight. The shape of the funnel, the social cadence underneath it, and how to build it without another app.
A scheduled action is a free chat tool firing the same prompt every Monday at 7am. It quietly replaced a habit tracker, a meal planner, and a weekly review app — and it raises a fair question about what those apps were really selling.
Population health data was built largely on male defaults, then handed to algorithms that smoothed every woman's outliers into “normal.” The fix is unromantic: treat your own body as a sample size of one, and read it with a model that does not assume you are the average of millions of strangers.
Most solo practitioners do not have an outreach problem. They have a Friday-afternoon problem. Here is the eight-channel system one coach installed in a weekend — and the small idea underneath it that any practice can copy.
Four wellness apps. Two wearables. Three diets. None of it stuck — until her practitioner stopped asking her to track and asked her to write four lines a week.
Most platforms teach a method and outsource their own back-end to someone else's stack. We built a recommendation engine using the same 3-Layer methodology we sell — and published the playbook.
The fifth food-logging app of the year is not the answer. The answer is one document, one model, and the willingness to stop outsourcing the question of what you actually ate.
Most small clinics, indie practitioners and solo founders have an outreach list and no outreach system. The list is a spreadsheet someone updates when they remember. The system is a Friday afternoon when no one has any energy left. Here is what changes when you wire one quiet board instead.
“Eat more protein. Lift heavy. Sleep eight hours.” Everyone already knows. The point of an individual AI stack isn’t to repeat the basics — it’s to walk into every appointment with sharper questions than the practitioner expected.
Every LLM company is racing to ship you an AI health product. The free general-purpose tools are quietly more useful — if you know how to use them. Health and wellness shouldn't be reserved for people who can stack ten subscriptions.
Semaglutide and tirzepatide change four signals at once. Here is how to read them yourself, with AI, instead of trusting the prescriber's dashboard.
AI is the most powerful instrument wellness has ever had. If you can read your own body, you can learn to direct this tool — and the possibilities open fast.
The honest case for using general-purpose AI as your health intelligence layer — instead of paying another subscription to read your own data back to you.
BPC-157, TB-500, ipamorelin, CJC-1295. The marketing is enormous; the human evidence is mostly thin. Here is the calm, honest reading.
Your skin is a slow signal. The feed is fast. Here is how to read your face like data and stop buying the next thing the algorithm sells you.
Why the things that matter for longevity move on the scale of years — and what to track if you actually want them to bend.
The plain-English explainer of the term — what it means, what it isn't, and why it doesn't need to be a product.
Some practitioners are being replaced — not by AI, but by other practitioners who use it better. The honest playbook for the AI-informed practice.
Daily perfection vs sustainable rhythm — what insight density actually looks like.
How to read your bloodwork through clinical, functional, and personal lenses without ideology.
Where the AI Health Stack ends, and where a human practitioner has to begin.
Why we chose Perplexity for research, Gemini for ledger, ChatGPT for protocol — and what changes if that breaks.
Building a personal HealthOS that survives version changes, deprecations, and acquisitions.
Strong, promising, anecdotal — labelled out loud. The three-tier hierarchy that prevents both naïve scientism and uncritical wellness.
Daily newsletters condition you to skim. Weekly letters earn the reading. The math on attention, signal, and the cost of being in someone's inbox.
New essays roughly once a fortnight. Subscribe in the footer for the Sunday letter.
Editorial principles
Every claim ranks itself: strong, promising, anecdotal. We label it. We don't hide it.
If we can't explain why something works, we don't recommend it. Even when it's trending.
One essay a fortnight, written until it's right. No publishing calendar pressure.
Templates and prompts are yours to copy, fork, modify. We benefit when you build, not when you stay.
No countdown timers, no 'last chance' subject lines. The Sunday letter arrives Sunday.
When something we recommended doesn't hold up, we publish the retraction at the top of the original.
The Sunday Letter
The week's essay, plus three signal-over-noise links from what we read. No tracking pixels. No re-engagement campaigns. Unsubscribe in one click.
Same list as the footer signup. One-click unsubscribe in every email. EU-built, GDPR-first.
Related
Pick the door that matches where you are. The Atlas points you at tools, the courses teach the method, the case studies show the work.
100+ AI tools sorted by what they actually do for your health stack — research, ledger, protocol. Updated quarterly.
Get the AtlasBi-weekly Zoom workshop with Sabin. Build your AI Health Stack end-to-end, ask one real question, leave with a working setup.
Reserve a seatBuild your own AI Health Stack in 4 weeks. Same method we use with private clients — Research, Ledger, Protocol.
See the coursesPersonalised
Based on what you've been reading — always learning.