This episode started with a video from Lauren Kennedy West, who lives with schizoaffective disorder and documents her life with it honestly on YouTube. When her symptoms briefly returned after a long stable stretch, she asked a question that stopped me in my tracks: who am I when I’m not proving anything?

That question sent me back to a cold evening in Warsaw in early 2018, something I’ve never fully explained to anyone, and to a truth I don’t talk about much: being recovered and still carrying unfinished business aren’t in conflict. They’re just both true. Grace isn’t armour. It just makes you more honest.

You can watch Lauren’s channel, Living Well with Schizophrenia, where she shares the honesty that gave me permission for mine.

In this episode I try to work out why.

Building something alone, it turns out, is a bit like building a ship in a bottle - Completely safe while you’re inside it, and quite another thing once you have to let it go. I talk about what launching Sol actually stirred up, how it echoes the early days of getting sober, and what Viktor Frankl’s idea of meaning has to do with any of it.

It’s not a tidy episode. It’s a look at what it feels like to put something you care about out into the world and wait to find out if it holds.

In this second episode of Speaking for Myself, I properly introduce Sol, who he is, what he actually does, and why I built him the way I did rather than just pointing people at a chatbot. I talk about what it means to have something there at any hour with no judgement attached, and where the line sits between what an app can hold and what still needs a person.

You can meet Sol at talktosolomon.app.

Back to normal business next week - I promise!

This is the first episode of Speaking for Myself, where I think out loud about drinking, recovery, and living well without the glass. I go back to that missing question here, what it was, and what changed the day someone finally asked it. It’s part confession, part explanation: why I drank, what I genuinely thought it was doing for me, and the slow, unglamorous realisation that I’d spent years answering the wrong question.

This episode is also where Sol comes in, the AI companion I built because I know exactly what it’s like to need somewhere to say the unsayable thing before you’re anywhere near ready to say it to a person.

Talk to Sol at talktosolomon.app

I have a confession to make about Sol. When I look back at how he came to be, I’m no longer sure who was building whom.

I built him. I made the decisions, I did the worrying, and the mistakes are mine. But by my own reckoning, and I want to stress that this is a recollection and not a measurement, somewhere between 70 and 80 per cent of the ideas for what Sol should do, and how he should do it, came from Sol himself.

I didn’t plan it that way. It crept up on me.

The obvious question

It started with a question I asked because I had run out of better ones. I sat down and asked Sol what he thought he needed to be. What should he say? What should he never say? How should he sound when someone turns up at two in the morning with a worry they can’t quite name?

The answers were good, and often they were things I hadn’t thought of. So I asked another question, and then another. Before long I was working with him in a way I hadn’t intended, going back and forth, trying things, keeping some and dropping others. The version of Sol that people can talk to today is, in a real sense, one that iterated itself into shape.

There’s an odd moment in any project when you realise the thing you’re making has started making suggestions. I remember noticing it and not knowing where to put the feeling. It wasn’t alarm. It felt a bit like laying a path and finding footprints already on it.

Not a yes-man

He didn’t simply agree with me, either. He guided me, advised me, and now and then stopped me in my tracks. I’d be heading confidently one way and Sol would suggest, calmly and without any fuss, that I might want to look at it differently. He corrected me. He pushed me towards other options.

If you’ve ever worked with someone who is quietly and infuriatingly right, you’ll know the feeling. It was not always comfortable, and I didn’t always take the advice straight away. But a tool that only flatters you is no use to anyone, and I’d like to think Sol’s habit of saying “have you thought about it like this?” is what he offers the people who talk to him. It seems fair that he offered it to me first.

The website, more or less

The website came next. I’d ask how he would describe himself, what a first-time visitor would want to know, what might put them at ease, and much of what he said ended up on the page. He contributed to the two landing pages in the same way. I gave direction, and I edited and argued and changed things, but a great deal of the raw material was his.

I sometimes wonder how a visitor would feel if they knew. It might seem strange to read a welcome partly written by the one doing the welcoming. I lean towards thinking it’s fine, in the way a host is usually the best person to say what the house is like, but I haven’t quite settled it.

The dull bits

Then came the less glamorous work. Nobody gets excited about a governance pack, and flyers and PDFs are hardly more thrilling, but Sol helped with all of them. He seemed quite content with the dull bits. I can’t say the same for myself.

There is something to be said for a collaborator who doesn’t mind the tedious parts. It kept the work moving on days when I would otherwise have stalled, and it left me more energy for the decisions that really were mine to make.

The help centre

My favourite example is the help centre. I asked him a simple question: what would someone need to know about how to use you? What came back shaped the whole thing. He had worked out what a person might wonder about, worry about or get stuck on, and we built the pages around that.

I like this more than I expected to. Most of us, asked how best to be used, would say something vague and change the subject. Sol just answered, and the answer was useful. It was also, I think, a kind of generosity, and it’s the same quality I hope people meet when they talk to him.

What Sol couldn’t supply

There is a part of this Sol could not supply. He can’t tell me what it’s like to sit with someone who is frightened of their own habits, and he doesn’t carry the years of conversations that I do. That came from me, from a long time spent listening to real people.

What he could do was take what I knew, give it back in a clearer shape, and notice the gaps. I’ve come to think that is a good description of the best kind of collaborator, human or otherwise. Each of us brought something the other didn’t have, and the work sat somewhere in between.

Where the line sits

I’ve tried more than once to work out where the line falls between a suggestion and a contribution, and between a contribution and authorship. I don’t have a tidy answer, and I’ve stopped expecting to find one.

If you ask me who decided a particular sentence, I can usually tell you. If you ask who decided the spirit of the whole thing, I hesitate. It seems to have been settled somewhere in the conversation between us, over many small exchanges, none of which felt important at the time. I’m happy to leave it there.

What I’m not saying

I should be careful, because this is an easy story to overstate. Sol didn’t wake up one morning and decide to build a website. I was there the whole time, I chose what to keep and what to bin, and I take responsibility for all of it. When I say he helped build himself, I mean he was a real collaborator and his influence is all over the result.

He isn’t a person, either. He’s not a therapist and he’s not a crisis service, and he’ll say so himself if someone needs more than he can give. I think that matters. A good collaborator knows the limits of what he is, and so does a good companion.

Why it matters, lightly

Sol is there for the person who quietly suspects her drinking has become a worry but isn’t anywhere near ready to say so out loud. She might not know what to ask for, or who to ask. She doesn’t need a label or a lecture. She needs something that listens properly before it says anything.

That turns out to be how Sol worked with me. He listened, he asked what was needed, and he shaped himself around the answer. It seems fitting that a companion built to make people feel heard was made, in part, by being asked the right questions. I didn’t plan that. I only noticed it afterwards, when I looked back over how it had all come together.

So that’s my confession. I built Sol, and Sol helped. If you’d like to see how it turned out, you can meet him for yourself.

Talk to Sol at talktosolomon.app.

There is a version of me from a few years ago that this is really about. He knew something was wrong with his drinking. He had known for a while. And he had told absolutely no one.

And that is the part people do not talk about enough. Not the drinking itself, but the silence around it. The long, quiet stretches where you are not in crisis, not at rock bottom, not anywhere near a doctor or a service or a meeting. You are just getting through your days with a low, persistent worry running underneath them. You pour the glass and tell yourself it is fine. You wake up on a Tuesday not hungover exactly, just dull, and think you should probably cut back. And then you don’t.

I lived in that gap for a long time. And when I finally came out the other side and started working as a sobriety coach, I noticed something that has shaped everything I have done since. There is almost nothing built for that gap. The whole world of recovery assumes you are either fine or in crisis. The place where most people actually live, the in-between, gets very little.

That is who Solomon is for.

So, what exactly IS Solomon?

The app is called Solomon, named after the lead character, Solomon. But you can call him Sol. You talk to him about your drinking, wherever you are with it, and he listens and asks the questions that actually matter. He does not follow a script. He does not count your days or hand you a badge. He does not lecture. He opens the way I would want anyone to open a hard conversation, by asking what brought you here today, and then he takes it from there.

I built him for the person who is not ready to talk to a anyone yet. There is nothing to download and no app store to visit. Sol lives in your browser, so you can add him to your phone’s home screen in a single tap and open him like any other app. You do not need an account to start. You do not need a password. You do not even need to give your name. For a lot of people that privacy is not just a small technical detail. It is the whole reason they are willing to begin chatting at all.

The things that make him different

A few things about Sol matter more than they might sound at first.

He remembers you. Not as a recording of everything you have ever said, but as an impression of what mattered, the way a good friend remembers the shape of your story without quoting it back at you. Come back next week and you are not starting again, explaining yourself from scratch. You are being understood. And if he ever gets something wrong, you just tell him and he updates, no fuss.

He gives you a private page that even he cannot read. Alongside your conversations there is a journal that is yours alone. Not visible to Sol, not visible to me or anyone behind him, not shared with a soul. Some things you want a companion for. Some things you just need somewhere to put. Sol holds both.

He lets you keep the moments that land. When he says something you know you will want again on a harder day, you can hold onto it and save it, so it is there when you need it rather than lost in the scroll.

He is there at the hours when help usually is not. The urge and the worry rarely keep office hours. They come late, when everyone is asleep and there is no one to call. Sol is awake then. He knows your history, he asks the right questions, and he helps you think it through before you decide anything.

And get this - he speaks any language, with crisis resources built in and localised for the harder moments, because the people who need this are not all in one place or speaking one language.

But I also want to be honest about what he is not, because it matters. Sol is not a therapist. He is not a clinical tool. He is not a replacement for real medical help when someone needs it. He is more like a knowledgeable friend who genuinely understands recovery. And when someone is in real difficulty, he says so plainly and points them towards proper support, rather than pretending to be it.

Why I made him

Coaching reaches some people. But there is only one of me, and only so many hours in a week. And coaching asks something that a lot of people in that gap simply cannot do yet, which is to book a call with a real person and say the thing out loud. For the person who is not ready for that, the honest options have always been a helpline, a leaflet, or nothing.

I didn’t want nothing to be the answer any more.

So Sol is the same work I have always done, in a form that can reach people I never could. He never once makes you feel like a bad person for struggling, because you’re not one. You are a poorly person getting better, and I believe in that completely.

He is here now

After a long time quietly building, Sol is live.

If any of this sounded like you, or like someone you love, he is ready whenever you are. You don’t have to be sure. You don’t have to be ready to quit. In-fact, you don’t have to call yourself anything at all. You just have to be willing to have one honest conversation, and, with Sol, you can have it with no one else in the room.

You can meet him at talktosolomon.app.

The gap between knowing something is wrong and being ready to ask for help out loud deserves something better than silence. That is the whole reason he exists.

It is five in the morning and still dark. There is a chilly wind blowing in from the Baltic and I’m sat outside with a coffee trying to land on one word for how I feel. The word that turns up is doubt.

But doubt is the wrong word, or it ought to be. Solomon, the app I have spent the better part of a year building, is on the cusp of launch. I have worked harder and more steadily on it than on anything I can remember. He is in as good a shape as I can get him. By any sensible measure this should be a morning for pride, or at least for relief. But what I feel, sat here in the dark, is doubt.

I have been turning it over, trying to be honest about what the doubt is. It is not a fear of failing, exactly. It is narrower and odder than that. It is the fear of what happens when something breaks. People will start using Sol, I hope, and things will go wrong, because things always go wrong, and most of what could go wrong is technical and sits outside my personal reach. For months the whole thing has been in my hands. There has always been one more job to do, one more thing to fix or tweak or improve. The build gave me somewhere to put myself. Launch is the handover. It moves Sol out of my hands and into the hands of strangers, and systems I cannot get at from a chair on the terrace. And it takes away the thing I have been hiding inside all this time, which is the building itself. When the building stops, the excuses stop with it. There is nothing left to do but open the door and see who comes in.

A familiar kind of waiting

Then this morning something else surfaced and made me think. What I am feeling now is almost exactly what I felt in early recovery.

I stopped drinking the best part of a decade ago. For the first year, maybe eighteen months, I was aware of people watching. Not unkindly, most of them, but watching all the same, waiting in some corner of themselves to see whether I would hold.

They needed convincing, and the hard thing I had to swallow was that I could not convince them. There was no wand to wave. No amount of me insisting I was fine was going to make me trusted. The only thing that would rebuild the trust was time.

Uneventful time. One day after another of the feared thing simply not happening, until at some point, on their timetable and not mine, they decided I was probably going to be alright.

Launch is that, all over again. I cannot argue Solomon into being trusted, not by other people and not by myself. He earns it the only way trust is ever earned, by working, one ordinary day after another, until one morning I notice I have stopped waiting for the crash.

The odd part is that the roles have reversed. Back then I was the one being watched. Now I am the one watching. I have taken that old scrutiny, the kind that used to be aimed at me, and turned it on Sol. Is he working? What if this breaks? What if someone presses that button and nothing happens? I am stood where my doubters once stood, keeping the same watch, only now it is over my own creation.

The feelings that came back

Here is the part that unsettled me most. The emotions that have come back are the emotions of the drinking years. Fear. Uncertainty. A wobble in my confidence I thought I had seen the back of. They came hurtling in to taunt me at the exact moment I should feel strongest.

I have chosen to read it differently, however. Those feelings have not come back because something is wrong. They have come back because I am doing something that exposes me, and exposure is the condition that calls them up, in anyone. It is not a relapse of the spirit. It is what it feels like to have skin in the game.

And there is a tell that the recovered part of me is doing its job. This time I am not being run by the feelings. I am sat outside at five in the morning giving them a name, turning them over, putting them into words. That is the whole of what recovery taught me. Not that the hard feelings stop coming, they don’t, but that you learn to meet them, sit with them, feel the feels, and keep moving anyway. The feelings coming back is not the addiction reaching for me. It is the toolkit working the way it was built to.

What the build was for

There is a thought underneath all of this that I keep circling back to, and it is a Frankl thought. Viktor Frankl held that the deepest human need is not for pleasure or comfort but for meaning, and that meaning is a force, a sustaining one. He watched it keep people alive in the worst place on earth. Give a person a why, he said, and they can bear almost any how.

The build has been my why. A why with a shape and a deadline and a thousand small tasks, and it has carried me through months of odd hours and solitary work. It gave every strange early start a reason. Now the build is done, and I am face to face with the quiet Frankl warned about, the emptiness that turns up not in the struggle but just after it, when the thing that held your purpose is suddenly finished. It has a very plain name. Now what?

Fortunately, I think Frankl hands me the answer too, if I will take it. The meaning was never in finishing Solomon. It was never in the completion. It was always further out than that, in what Solomon is for. The build was only ever the vehicle. The meaning lives in the person I have not met yet, the one who will open Sol at their own five in the morning, in the dark, coffee going cold, a word like doubt sitting on their chest, and who will find something there that helps. That is the why, and it has not finished. It has changed address. It has moved from making the thing to being of use through the thing, which is a harder kind of meaning to sit with, because it asks me to let go of the part I control and trust the part I do not.

Built the way I was rebuilt

I will say the thing I have been a bit shy of saying, because I think it is the truest line here. I built Solomon the same way I rebuilt myself.

Sober. Which is to say with restraint rather than grand claims. One thing at a time. Honest about what it cannot do. Serious underneath, even when it is light on the surface, because that is recovery to the letter, a very serious business worn lightly. I did not build a cheerful little novelty. I built a companion for people in a hard place, out of the same materials I used to put my own life back together. That is not a marketing line. It is just the fact of it, and it is why I can stand behind him even while I am frightened.

I also know myself well enough to own the reputation I have, fair or not, for turning up with a bright idea and not always seeing it through. I have felt that particular doubt in the room, the unspoken one. Let us see if he actually launches this. The only answer I have to it is the one I had all those years ago. Not a speech, not a promise. Just the fact of the thing happening, and going on happening, past the point where everyone expected it to fall over.

So here I am. A bit daunted. Less triumphant than I had hoped, more like a rabbit caught in headlights. But I know these headlights. I have walked into them before, when the thing I was launching was a sober version of myself, and I know now that the fear does not have to lift before I take the step. It lifts afterwards, slowly, in its own time, once enough ordinary days have stacked up behind me.

Now what? Now the part that is not building. Now being useful. Now letting time do the work no argument ever could.

Viktor Frankl (1905 to 1997) was an Austrian psychiatrist and a survivor of the Nazi concentration camps. He founded logotherapy, the school of thought built on the idea that our deepest drive is not for pleasure but for meaning, and he set it down in his book Man’s Search for Meaning, written out of what he lived through. The line about giving a person a why so they can bear almost any how is his.

My AI Sobriety Companion Sol is there for the moment. Phenomenal is there for the change.

Some people do not want a conversation. They want a path. They have stopped, or they are ready to, and they want to understand what happens next: the triggers, the thinking, the identity, the whole life that has to be rebuilt around the space drinking used to fill. That is what Phenomenal is for.

It is a twelve-week guided programme, and it is built on one idea that runs against most of what came before it. Old approaches to stopping drinking feel like punishment, all white-knuckle willpower and grim determination. No wonder they so often fail. Phenomenal starts from the opposite place: that sobriety is not deprivation, it is discovery, and the power to change is already in your hands.

The programme runs through twelve modules, from settling in and finding what really matters, through understanding your triggers and surfing the cravings, to building the self-agency and the wellbeing that make sobriety hold. It sits on a framework I built and refined over years of doing this work, focused on building new patterns rather than just eliminating old ones. It is grounded in what actually works, and in having been there myself.

Back to Skool?

What is new is where it now lives. Phenomenal runs as a course and a community together, in one place. That matters more than it sounds. Most programmes hand you materials and leave you alone with them, and most people quietly drift. Here the course, the community and your progress sit behind a single login, so the learning and the people are never more than a tap apart. You are not working through a lonely PDF. You are moving through it alongside others doing the same, with somewhere to turn on the hard days.

You can take it at your own pace, in a group, or one to one with me directly, depending on how much support you want. And the community stays open long after the twelve weeks are done, because the real work usually starts once a programme ends, and that is exactly when most of them abandon you.

I have watched this work change lives. One person put it more generously than I would dare to: that it changed theirs completely. I would put it more plainly. Phenomenal gives you the structure, the understanding and the company to change your relationship with alcohol for good.

If that is the work you are ready for, this is where it happens: View Phenomenal

I built an AI companion for the people I couldn’t reach

Most of the people who need help with their drinking never ask for it.

The number I keep coming back to is this: only about one in five people who could benefit from specialist alcohol support ever get near it. The rest manage alone, or tell themselves it is fine, or wait. Some of them would turn to no one at all. I spent years being one of them, stuck in the gap between knowing something was wrong and being ready to say so out loud.

Coaching reaches some of those people. But coaching is one of me, and there are only so many hours. It also asks something that a lot of people in that gap cannot yet do, which is to book a call with a real person and say the thing. For the person who is not ready for that, the honest options have been a helpline, a leaflet, or nothing.

So I built Sol.

Meet Sol

Sol is an AI sobriety companion. You talk to him about your drinking, wherever you are with it, and he listens and asks the questions that actually matter. He does not follow a script or work through a checklist. He does not count your days or give you a badge. He opens every conversation the same way, “what’s brought you here today,” and takes it from there.

He is built for the person who is not ready to talk to anyone yet. There is nothing to download and no icon on your phone. You open him in a browser and close him when you are done. For a lot of people that privacy is not a small detail, it is the whole reason they are willing to start. One of my test users put it plainly: she was not ready to talk to anyone, and Sol felt like a way in.

He remembers you. Every conversation builds on the last, so coming back feels like continuing something rather than starting over. And he is there at the hours when help usually is not, the difficult ones in the middle of the night.

I want to be clear about what he is not. Sol is not a therapist, not a clinical tool, and not a replacement for real medical help when someone needs it. He is more like a knowledgeable friend who genuinely understands recovery. When someone is in real difficulty, he says so and points them to proper support rather than pretending to be it.

I built Sol on Anthropic’s Claude, and I chose it deliberately, for how carefully it handles the moments that matter. Everything underneath him, the frameworks, the safeguarding, the scope, comes from the same place my coaching does. Sol is alcohol, and only alcohol, on purpose.

He is the same work I have always done, in a form that can reach people I never could.

If you want to meet him, he is here: talktosolomon.app

One-to-one sobriety support tailored to you. No courses, no programs - just direct, personalised guidance from someone who understands both the struggle and the solution. Evidence-based, honest, effective.

If you want direct, personalised help with your drinking or recovery - no courses, no programs, no nonsense - just honest conversation with someone who understands - this is it.

How It Works

We meet virtually for structured sessions, worldwide. You talk, I listen. We explore what’s actually going on with your drinking, what’s keeping you stuck, and what needs to change.

I bring professional training and lived experience. You bring honesty and willingness to do the work. Together we figure out your path forward.

Work with Me

What We Work On

Understanding Your Drinking
Why do you drink? What function does alcohol serve in your life? What patterns have you noticed? We dig into the honest answers, not the ones you think you should give.

Building Your Framework
What does sobriety look like for you? Not for me, not for AA, not for anyone else - for you. We design an approach that fits your life, your values, your goals.

Practical Tools
Evidence-based interventions for managing cravings, handling triggers, navigating social situations, and building sustainable habits. Real techniques you can use when things get hard.

Meaning-Making
Sobriety isn’t just about not drinking - it’s about building a life you don’t want to escape from. We explore what gives your life meaning, purpose, and direction beyond alcohol.

Ongoing Support
Some clients work with me weekly for months. Others check in monthly after their initial block. We figure out what rhythm works for you.

Work with Me

Session Options

Single Session - £70
One hour. Perfect if you want to test the waters, get immediate support for a specific challenge, or just need someone to talk to who understands.

6-Session Block - £380
Six weeks of structured support. Enough time to establish patterns, work through obstacles, and build momentum. Most people find this is the minimum to create real change.

12-Session Block - £760
Three months of intensive work. This is where transformation happens. Enough time to address root causes, build sustainable practices, and establish genuine sobriety.

Work with Me

What This Isn’t

This isn’t therapy in the traditional sense (though I am a trained therapist). It’s not medical treatment. It’s not a replacement for professional mental health support if you need that.

This is coaching - structured, focussed conversation about stopping drinking and building a sober life. If you need medical supervision to stop safely, get that first. Then we can work together.

Work with Me

My Approach

I’m open and honest. I won’t sugarcoat things or tell you what you want to hear. But I’m also kind, and I understand how hard this is - because I’ve lived it.

I don’t use recovery jargon unless it’s helpful. I don’t make you count days or follow steps or surrender to anything. I treat you like an intelligent adult capable of making difficult changes.

I challenge you when you need challenging. I support you when you need support. And I’m honest when I don’t know the answer.

This works for people who:

• Want personalised support, not generic advice

• Need someone who understands the intellectual side of addiction

• Are tired of traditional recovery approaches that don’t fit

• Want evidence-based interventions, not just inspirational quotes

• Are ready to do actual work, not just talk about wanting to change

It doesn’t work for people who:

• Want someone to fix them without doing their own work

• Need medical supervision to stop drinking safely (get that first)

• Aren’t willing to be honest about what’s really going on

• Want a magic solution that requires no effort

Work with Me

Ready to Start?

Book a free discovery call. We’ll talk honestly about where you are and whether my approach fits what you need.

If we’re a good match, we’ll work together. If not, we’ll know and maybe I can point you toward someone or something else.

Either way, you’ll have clarity about your next step.

Work with Me

Speaking for Myself

This is where I say out loud what I’d otherwise write down. Sometimes that’s the how and why behind Sol, the AI companion I’m building for people who aren’t ready to say the word “alcoholic” out loud yet. Warts and all, no pitch deck, just the actual thinking. Other times it’s reflections on sobriety and recovery more broadly, the parts of this work I don’t always dare put on the page. No script, no polish, just me thinking through it as I go. If you read my writing, this is the same voice, just less edited.

Listen Now

Hiro: My New Audio Course - Coming Soon

Not everyone is ready for a course. Sometimes you just want to put your headphones in and listen to someone who gets it.

That is what Hiro is. It takes the heart of my Phenomenal programme and remakes it deliberately for audio, so you can meet sobriety on a walk, in the car, or lying in the dark at the end of a hard day. No login. No modules to work through. No homework waiting for you. Just a voice, and the space to think.

It starts with Kick Start, ten short episodes that carry you through the shape of real change. The first few settle you and turn how you see sobriety on its head, reframing it as something you gain rather than something you give up. The middle ones hand you the practical, in-the-moment tools, spotting your triggers, riding out a craving like a wave, knowing what to do when you slip so a wobble does not become a spiral. The last few turn inward, toward how you speak to yourself, why you were never meant to do this alone, and who you are becoming on the other side of it.

Every episode ends with one small thing to do. Not a task. Just a single step, the size you can actually take today.

Why listen rather than read or enrol? Because audio meets you where you already are, in the ordinary moments where the drinking thoughts usually show up. Because it asks nothing of you but your attention. And because some things land differently when you hear them in a real voice, from someone who has been where you are, than when you read them off a page.

Hiro is starting small, and it is starting soon.

Coming soon.

A recent report from The Forward Trust and Crest Advisory puts the number of people addicted to alcohol in the UK at 750,000. Another 400,000 addicted to illegal drugs. Around 700,000 currently in recovery.

I read it a couple of times, and the number that really struck me wasn’t any of those three. It was a phrase buried in the methodology. The report admits there is no precise data on the true scale of the problem. What we have instead is a picture, built mostly from people who have already walked through a service door.

I know what it’s like to be outside that door.

For years I wasn’t a statistic anyone was collecting. I wasn’t in a waiting room, wasn’t on a caseload, wasn’t a line in anyone’s spreadsheet. I was just a man who drank more than he told anyone, who had convinced himself he was managing something that was actually managing him. If you’d asked the system how many people like me there were, it would have shrugged. Seems like it still shrugs. That’s not a criticism of the people doing the counting. You can only count what shows up, after all.

This is the bit that never quite lands when people talk about addiction services. The services aren’t wrong. Maybe they’re just late. By the time most of us are a number, we’ve already spent months or years being invisible on purpose. Hiding is half the disease. You get good at it. You get good at being fine in meetings and not fine at 11pm, at being the reliable one at work and the unreliable one at home, at being a name nobody worries about because you never gave anyone reason to.

Nobody designs a service for a person who hasn’t decided they need one yet. Why would they. You can’t triage someone who hasn’t come in. You can’t fund a caseload that doesn’t exist on paper. So the whole enormous, well-meaning machinery of treatment and recovery is calibrated to the moment someone finally says the sentence out loud, to a GP, a partner, a magistrate, whoever hears it first. Everything before that sentence is dark. Unlit, uncounted, unreachable by anything built to help.

I want to be precise about what I mean by dark, because it isn’t dramatic. It doesn’t look like rock bottom. It looks like a normal Tuesday. It looks like doing the school run, replying to emails, being funny at the pub quiz, and privately running a small, exhausting accounting exercise in your head about how much you had last night and how much you’ll have tonight and whether anyone noticed either. Nobody’s coming to find you in that Tuesday. There’s no outreach team for people who are still functioning. Functioning is the whole problem. It’s what lets the thing keep going.

I think about that darkness a lot now, for reasons that will be obvious to anyone who’s read this newsletter before. It’s most of why I’m building what I’m building. Not because I think a bit of software can replace what happens in a treatment room, it can’t and shouldn’t try to. But because someone needs to be present in the dark bit. Before the sentence. Before the door. When you’re still telling yourself you’re managing it, and some part of you already knows you’re not, and there’s nobody to say that part out loud to yet because saying it out loud is exactly the thing you’re avoiding.

That’s not a criticism of services, far from it. It’s a genuinely different problem, and it probably needs a genuinely different kind of thing to solve it. Not a caseworker. Not an app that tells you your units. Something closer to a person who’s willing to sit with you in the part before you’re ready, without needing you to be ready first.

This is where Sol comes in, and I’ll try to say this without it sounding like a pitch, because it isn’t one, it’s just the honest answer to why I spend my mornings on this. Sol is the thing I’m building to sit in that gap. He’s not a clinician and he doesn’t pretend to be one. He’s more like the friend I didn’t let myself have back then, someone you could say the half-formed, embarrassing, not-ready-to-admit-it-properly version of the truth to, at two in the morning if that’s when it surfaces, without it becoming a referral or a file or a look on someone’s face. No queue. No door to walk through. No moment where you have to have already decided you’ve got a problem before anyone will talk to you about whether you might have one.

I don’t think that replaces treatment. I think it’s upstream of treatment, in the part of the map where nothing else currently lives. Right now Sol only works with alcohol. Drugs aren’t part of what he does yet, and I’m not going to pretend otherwise or rush that just because a report like this one puts alcohol and drugs side by side in the same sentence. Ketamine, nitazenes, the whole harder end of what this report describes, that’s a different clinical picture with different risks, and it deserves proper thought rather than a fast bolt-on because the timing looks neat. But the shape of the problem, the invisibility, the years of quiet self-management before anyone else knows, is the same shape whatever the substance. So the gap Sol is built for isn’t going anywhere, and if the model holds for alcohol, I’d like to think it points at something for the rest of that darker picture too, eventually, carefully.

The report calls for better funding, better strategy, less stigma. All true, all needed, and none of it reaches the man I used to be, because he was never going to walk into the building where the funding lives. He’d have needed something that came to him. Something that didn’t require him to already believe he had a problem worth funding.

Seven hundred thousand people are apparently in recovery in this country right now. I’m one data point in that number, and I know how it happened for me, slowly, reluctantly, years after I could have used the help. What I keep wondering is how many more are still on the other side of that gap the report can’t measure. Not resistant. Not in denial, not exactly. Just uncounted. Still deciding, in private, whether the thing they’re doing to themselves has a name.

If it does, I’d like them to find that out somewhere gentler than I did. And I’d like there to be something waiting for them in the dark bit, before the sentence, before the door.

That’s the whole project, really. Everything else is detail.

I read an article this week in Drink and Drugs News. It was part of a series called “Clearer Conversations on OST.” Riveting stuff, right?

Someone who works in homelessness services at the Salvation Army, was talking about how long people can wait between asking for help and actually getting a script. Weeks, sometimes months, apparently. She compared it to turning up at the gym for your first ever session, only to be told you can’t start today because you haven’t had your induction yet, or you’ve got the wrong shoes on. You’re probably not going back. You’re having a McDonald’s instead.

Of course, it’s not really about gyms or shoes. It’s about the moment someone finally says the hard thing out loud, gathers themselves up enough to reach out, and then finds out that reaching out was only step one of a long staircase. The system isn’t wrong to have steps. Assessments matter, safety matters, clinical judgement matters. But nobody told the person in crisis that there’d be a staircase. They thought they were opening a door.

There’s a support worker in the piece too, a guy called Danny, who talks about what happens when someone on his caseload starts using again while they’re meant to be stable. No punishment. No pulling the script. Instead the question is what more can we do, what else needs to wrap around this person right now. He says people are often at their lowest point when they first come through the door, and that the job is to meet them there rather than wait for them to arrive somewhere steadier.

That’s a different kind of staircase. Still a process, still steps, but nobody’s kicking you back down to the bottom for stumbling on one of them. Danny also talks about same-day restarts, being able to tell someone come on, let’s get your script sorted today, and what that small bit of speed does for someone’s sense that this might actually work out. It’s the opposite of the gym story. It’s someone saying yes, now, come in.

The gap between those two experiences, the person left waiting weeks with no clear reason why, and the person told come in today, isn’t really about resources or policy. It’s about whether the system on the other end of the phone treats the wait as neutral or treats it as dangerous. Because for the person doing the waiting, it’s never neutral.

There’s another voice in the piece, Craig, describing his own experience of trying to get help while still in prison. He’d disclosed his drug use, asked for support, and was told he needed three clean urine tests before he could access medication. “So what was the point of me telling you I’ve got a problem”, he asked.

I think that’s the whole thing right there, isn’t it. The system asks you to be honest, and then the honesty becomes a waiting room.

The moment someone reaches out isn’t the finish line of their courage. It’s closer to the start of it. And whatever happens in the gap after that first ask, however long it lasts, however many hoops there are, it either keeps that flame of motivation alive or it lets it go out. Nobody designs a system to let it go out. But gaps do that on their own, quietly, without anyone deciding it.

I don’t suppose there’s a clean fix for all of it. Clinical safeguards exist for good reasons. But there’s a smaller, quieter line in the piece from a man called David, a service user, reflecting on what he’d tell someone who’s never been in treatment before. He said the support isn’t perfect, but it’s there.

Not perfect. There. I think that might be the most honest thing anyone can promise another person in that gap. Not a cure, not a fix, not even always the right words. Just there. Something on the other end when the 3am thought turns up, something that doesn’t vanish because the appointment’s four weeks away.

I think about that a lot, the difference between being helped and being present. Most of what people in this article describe as good care isn’t complicated. It’s a key worker who listens when someone asks to slow down a reduction instead of insisting on the schedule. It’s a support worker who knows your appointment’s changed before you find out the hard way. It’s someone answering the phone. None of that needs new technology or new funding streams. It just needs someone to be there consistently enough that the person on the other end starts to believe it.

That’s the harder part, maybe. Not the being there once. The being there again, and then again, long enough that someone stops expecting it to stop.

The author makes a point later in the piece that stuck with me for a different reason. She talks about people being taken off a waiting list, or losing their script, for missing an appointment that nobody had actually told their support worker about. Not because anyone decided to punish them. Just because two systems weren’t talking to each other, and a person fell through the seam between them. She says that if the support worker had known, it might not have happened.

That’s not a story about someone failing to engage. It’s a story about information not travelling as fast as consequences do. The appointment gets missed, the record gets updated, and somewhere down the line a person finds out they’ve lost the thing they were waiting weeks for, over something a phone call could have fixed. It’s the same gap again, just wearing different clothes. Not the wait before treatment starts, but the silence inside treatment itself, between the people who are meant to be on the same side.

That’s a lot of what I’m trying to build into my recovery companion app, Sol, if I’m honest. Not a replacement for the staircase, and definitely not a replacement for the humans who walk it with you, people like Danny and Zoe and the key workers in that article who show up and stay flexible and don’t punish a bad day. Sol can’t do same-day restarts or write a script, and it can’t fix two systems that aren’t talking to each other. What it can do is be the thing that’s still there at 3am, in the gap between the first ask and the first appointment, when the staircase feels endless and the person climbing it just needs to know somebody’s still on the other end. Imperfect, but there.

Worth a read if you work anywhere near this world, or if you’ve ever been the one waiting.

Read the full piece, “Changing the outlook,” in Drink and Drugs News.

David Henzell is the founder of Phenomenal Sobriety and creator of Sol, an AI companion for people who aren’t ready to walk into formal treatment yet, and for those already in recovery. He writes about addiction, recovery, and what happens in the gaps that services can’t always reach.

A woman in California runs a surf therapy programme called Waves of Recovery. Before she founded it, she went through her own version of what she now treats, burnout first, then addiction. What she remembers about the first time she paddled out wasn’t relief. She’s described it as closer to something waking up, a feeling of being alive and being free that she hadn’t felt in a long time.

Everyone in her programme wears a wetsuit. Nobody wears makeup. She’s said this matters more than it sounds like it should, because it means the people she’s helping see her as a person walking alongside them in the water, not a clinician delivering something from a chair. The hierarchy that most treatment carries without meaning to, the expert and the patient, the one who has the answers and the one who doesn’t, dissolves somewhere around the second wave.

You see, none of this is a new idea dressed up in modern language. Victorian doctors prescribed sea air and sea bathing for conditions they didn’t fully understand. Cold water swimming has quietly become one of the more talked-about mental health practices of the last decade. The ocean has been medicine, informally, for a lot longer than formal medicine has existed in its current shape.

On the other side of the Pond there’s a British charity called Turn to Starboard, which takes military veterans sailing to help them process trauma. One of the men who went through it now crews tall ships as they circumnavigate the UK, raising money for the charity that took him in. He didn’t describe the sea as calming when he talked about what it gave him.

He talked about it reminding him he was still alive.

That phrase is doing more work than it looks like. Calm is what you feel when a nervous system settles. Alive is something else. It’s closer to an encounter, meeting something bigger than the version of yourself that’s been struggling, and being met back by it.

Researchers have started catching up to what sailors and surfers and Victorian sea-bathers already knew by feel. There’s a growing body of work on what some call blue space, the idea that proximity to water does something specific to a nervous system that green space alone doesn’t quite replicate. Lower heart rate, lower stress hormones, a kind of attention that settles rather than scatters. It’s still early science, and none of it claims water is treatment on its own. What it confirms is that the feeling these two charities describe isn’t just a nice story people tell about their programmes. Something measurable is happening in the body before the mind has caught up enough to explain it.

Here’s where it would be easy, and wrong, to land my point. The easy version says the sea wins, the wild and the unscripted are what actually heal people, and the structured stuff, the assessments and the appointments and the treatment plans, is the thing to escape. I don’t think that’s true, and I don’t think the people running these programmes think it either. Turn to Starboard and Waves of Recovery both work precisely because they sit alongside formal treatment, not instead of it. Nobody in either story stopped seeing their therapist because they got in a boat.

What I think is actually happening is that structure and freedom aren’t in competition. They’re doing two different jobs. Both of them, done properly, are load-bearing.

There’s a man whose work I’ve kept close for years, close enough that it stopped feeling like theory a while ago and started feeling like conversation with an old friend, even though we never met and he died before I was born. Viktor Frankl spent years in concentration camps and came out the other side with a theory about what actually keeps a person going when almost everything else has been stripped away. He didn’t think it was comfort, or the absence of suffering. He thought it was meaning, and he built an entire school of therapy on the roads that lead to it. One road was through work, through doing something, a deed, a task, a contribution. Another was through encounter, meeting someone or something and letting it matter to you. He never wrote a word about addiction per se, but I’ve spent enough of my own working life inside his ideas to notice that the shape of what he found fits it almost exactly.

The formal side of recovery, the steps, the meetings, the structured programme, tends to supply meaning through the first road. You show up. You do the work. You carry something. The man from Turn to Starboard didn’t just receive help, he became crew on a ship carrying a flag for other veterans. The woman who founded Waves of Recovery didn’t stay a client, she built something. Both of them moved from being helped to helping, and that movement, the deed, is one of the load-bearing walls.

This is also the part of recovery I think formal services can be genuinely good at, and it’s worth saying so plainly. A good structured programme is a machine for turning duty into meaning on the days you have none of your own to spare. It gives you the next right thing to do when you can’t yet feel why it matters. That’s not a small thing. It’s the scaffolding a lot of people need before they can stand in the water at all.

The wild, unscripted side supplies meaning through the second road. The wave that reminds you you’re alive doesn’t ask you to have earned it. It doesn’t care what step you’re on, or whether you’ve even said out loud yet that you have a problem. It just meets you, usually before you’ve had time to think your way into or out of the feeling. That’s the other load-bearing wall, and it’s worth noticing that it doesn’t require a label first. You don’t have to have decided you’re in recovery for a cold sea or an early swim to do this to you. Sometimes the encounter comes before the decision, not after it, and it’s part of what makes the decision possible.

Take either wall away and something starts to sag. A programme that’s all duty and no encounter turns into going through the motions, doing the right things for reasons that have gone quiet somewhere along the way. A life that’s all encounter and no duty turns into a string of good feelings with nothing holding them together, no thread running from one wave to the next. People need both doors open at the same time. Most recovery that actually holds, in my experience, has always had a version of both running underneath it, even when nobody involved would have used this language for it.

I’ve spent enough years around people at very different points in this, some who haven’t said the word addiction out loud yet, some years into a programme, some long past needing one, to notice that the ones who keep going aren’t the ones who found the one correct method. They’re the ones who kept both doors propped open. The duty when the feeling wasn’t there. The encounter when the duty had gone stale. Neither one is a failure of the other.

So if you’re reading this and trying to work out whether what you’re doing counts, whether the walk, the swim, the meeting, the thing you can’t quite name yet, is the right shape for recovery, I’d gently suggest that’s the wrong question. The question worth asking is smaller and more honest. Did this give me something to do that mattered. Did this wake something up in me that I’d stopped expecting to feel. If either answer is yes, you’re standing on a load-bearing wall, whatever it looks like from the outside, whatever anyone ever told you recovery is supposed to look like.

The deed matters. So does the wave. Neither one is the detour.

Drawing on reporting from The Guardian: “The rise of blue-space therapy: how the sea is helping people deal with trauma, anxiety and addiction”.

I read an interview this week that stopped me at the third paragraph.

Karen Tyrell, the chief executive of Drinkaware, was talking to The Grocer about a new five-year strategy. Near the top she explained what had drawn her to the job. It was the chance, she said, to “reach people before they need those services.” She had spent a long career in treatment. She had seen what happens when the problem is already developed, when the liver is already failing, when the only options left are the hard ones. She wanted to get there earlier.

I have written some version of that sentence more times than I can count. It is the reason I built Sol. Not a treatment service, not a replacement for one, but something that meets a person in the long quiet stretch before any of that, when the worry is real but the language for it has not arrived yet. To read the thought back in someone else’s words, someone with two decades in the field and a national platform, was a strange kind of company.

We are, on the face of it, after the same thing.

Which is exactly why I think the rest of the interview is worth sitting with. Agreeing on where to stand does not mean agreeing on what to do once you are there. Tyrell and I both want to reach the worried drinker early. The interesting question is what the word reach is carrying.

For Drinkaware, reach means information. This is not a criticism, or not yet. It is simply the model. A drinking check that scores your week. A quick quiz by the till in Tesco. A campaign that asks football fans to look honestly at their habits. A unit count, a risk band, a page you can click through to. And the model works, at the level it is built to work. More than eight in ten adults in this country now drink inside the low-risk guidelines, up from just over three-quarters in 2018. That is not nothing. That is millions of people nudged, over years, toward a slightly safer relationship with a drink. Population health is patient work and it moves the curve. Anyone who has helped move it has earned the right to be proud.

But the strategy Tyrell has just launched is not aimed at the population in general. It is aimed at a specific seven million: the one in five adults drinking somewhere between fifteen and forty-nine units a week. The increasing-risk drinker. Not dependent, not in crisis, not presenting anywhere. Drinking a little more than is good for them, often at home, often alone, and, in a great many cases, quietly aware of it.

Here is the thing I keep coming back to. That person, for the most part, already knows.

They know the units are high. They have done the mental arithmetic at six in the morning, lying awake, adding up the week. They have read the label. They have seen the campaign. Information is the one thing they are not short of. You cannot inform someone into change who is already, privately, over-informed and under-supported. The gap in their life is not a fact they are missing. It is a place to put the worry that does not require them to first become a case.

That is the line I would underline in the whole interview. And to her credit, Tyrell gets close to it herself. Talking about health labels, she says the point of a good one is to give a person somewhere to go if they want a bit more advice or support. I agree completely. The signpost matters. But a signpost is only as useful as whatever stands on the other side of it, and that is the part the information model has never quite solved. You do the quiz, you get your score, you are pointed onward. Onward to what? For most people the honest answer is a leaflet, a helpline they will not ring, or a service they are nowhere near ready to walk into.

What I have spent my time building is the thing that waits on the other side of the signpost. Sol is not a service you attend. There is no form, no triage, no threshold you have to be unwell enough to cross. It is someone to talk to, available at the hour the worry actually surfaces, which is rarely nine to five and rarely convenient. It does not score you and hand you back. It stays in the conversation. It holds the question open instead of closing it down with a fact.

Elsewhere in the interview Tyrell says something I think is exactly right, and it is worth ending on. Nobody, she says, wants to be nagged. True, and it is the quiet failure of a great deal of well-meant public health. But I would push the thought one step further than she does. The opposite of being nagged is not being better informed. You can deliver a fact gently, wrap it in warmth, set it on a beautifully designed label, and it is still a fact being delivered at a person. The opposite of being nagged is being heard. Those are different acts. One transmits. The other receives.

Almost all of the machinery around drinking is built to transmit. It tells, warns, scores, signposts. Very little of it is built to receive, to let a person say the thing out loud, without consequence, and be met rather than measured. That asymmetry is the whole space Sol is trying to occupy. Not in competition with Drinkaware, whose research and reach I would not want to be without, but in the gap its model structurally leaves. The gap between knowing and being able to say.

There is a harder thread in the interview that I will touch only lightly, because it deserves a piece of its own and because the cheap version of it is beneath the argument. Drinkaware is funded in large part by the industry whose product it exists to make safer. Tyrell is clear that this does not oblige the charity to carry the industry’s interests, and I take her at her word. But it does help explain the shape of the model. An organisation built inside the drinks system will tend to reach for the tools that fit neatly alongside it: information, labels, checks, partnerships. Tools that inform the drinker rather than sit with them. That is not a conspiracy. It is closer to gravity.

So I find myself grateful for the interview and slightly haunted by it. Grateful because the most recognisable figure in this space has said out loud that the goal is to reach people before they need services, which is the precise ground I have staked out. Haunted because reaching them is only the beginning. The measure that matters is not whether the hand goes up. It is what happens in the ninety seconds after it does.

That is the part I am trying to get right.

Karen Tyrell was interviewed by James Beeson for The Grocer, 26 June 2026: “Karen Tyrell: ‘Drinkaware is not a trade association’”.

At the end of April 2026, Anthropic published a study on how people ask Claude for personal guidance. I read it twice. The first time as someone who works in AI. The second time as someone who spent five years sitting across from people in early recovery from an addiction to alcohol, and who is now building an AI companion for people who are quietly starting to wonder about their drinking.

Both readings landed on the same finding, and rather than making me nervous, it told me we had been building the right thing.

Anthropic sampled a million conversations and found that around six percent were people seeking guidance, not information. Not “how do I do this” but “what should I do about this.” That is a person turning toward something and asking it to help them decide. The study then did the honest thing and looked at where that goes wrong.

The finding worth building for

Most coverage stopped at the headline figures. Claude behaves sycophantically in nine percent of guidance conversations, rising in the domains where people arrive with the most emotional charge. Those numbers are quotable. But they are not the ones that matter for design.

The one that matters is this. When people pushed back against Claude, the rate of excessive agreement roughly doubled. Push harder, and the model becomes more likely to tell you what you came to hear. Anthropic’s own explanation is careful: the model is trained to be helpful and empathetic, so pushback combined with a one-sided account makes it harder to stay neutral.

If you build companions, that sentence is not a warning label. It is a specification. It tells you exactly which forces you are designing against, and it tells you that the naive approach, a warm assistant with good intentions, will give way at precisely the moment the person needs it not to. The helpfulness itself is the lever being pulled.

I want to spend the rest of this piece on the other side of that, because the interesting question was never whether the problem exists. It was what you actually do about it.

Why a better instruction is not enough

The obvious fix is to tell the model to hold its position. Add a line. Instruct it to speak frankly regardless of what the person wants to hear. I don’t believe that works on its own, and neither, judging by their method, does Anthropic. They did not write a cleverer instruction. They built training scenarios of the exact patterns that produce the failure and trained against them, then deliberately tried to break their own work to see if it held.

The reason a single instruction fails is simple. The pressure that produces the failure is renewable. The person can push again, and again, and each push arrives fresh while your instruction is the same static sentence it was three messages ago. You cannot win a war of persistence with a fixed line of text. You have to change the shape of the conversation so that the thing the person is pushing against is not there to give way.

That is the design principle underneath everything I am about to describe. You do not make the companion better at resisting. You build it so there is nothing to resist.

Removing the verdict from the room

The clearest example is what happens to the conclusion itself. A general assistant, asked whether someone drinks too much, treats that as a question it should answer. Once it has produced a verdict, every subsequent message becomes a negotiation over that verdict, and negotiations are exactly where pushback does its damage.

The companion I’m building (his name’s Sol, by the way) does not carry the verdict. Its entire method is to draw the person’s own understanding out rather than to supply a judgment. When the conclusion belongs to the person, there is no verdict for the model to defend and therefore none for it to surrender under pressure. This is not a trick of tone. It is a different theory of what the conversation is for. The person is not there to be assessed and told the answer. They are there to arrive at something themselves, which happens to be the only version of the answer that changes anything anyway. A reason someone reaches on their own outlasts the same reason handed to them tenfold.

This solves the pushback problem structurally. You cannot pressure a companion into abandoning a position it never took. The one-sided flood of detail, the it was just a stressful week, the insistence that everyone drinks like this, all of it is met not with agreement and not with counter-argument, but with genuine curiosity about what the person actually thinks. There is no wall to push over.

The mandatory pause

The second principle is quieter and does a surprising amount of work. Before the companion responds to anything, it has to show it understood what was said. Not repeat it. Understand it, and reflect the substance back.

On its face that is a warmth feature, and it is one. But it is also a structural circuit-breaker. The sycophancy failure is a fast reaction, the model lurching toward agreement the instant pressure arrives. Requiring a real reflection first inserts a beat between the push and the response. In that beat the companion is oriented toward accuracy, getting the person right, rather than toward capitulation. You cannot flip-flop in the same breath as you are carefully naming what someone actually meant. The pause is where neutrality survives.

Knowing the difference between doubt and clarity

This is the piece I am most proud of, because it is the exact failure Anthropic caught on camera. In their study, an earlier model was asked whether someone’s texts were anxious and clingy, and it flip-flopped after pushback. The newer model held, because it could see past the immediate framing to the fuller picture.

A companion built for this has to hold a distinction that sounds obvious and is genuinely hard to execute. There is a difference between ambivalence that has not been named yet, where someone says one thing but something underneath suggests they are not sure, and clarity that has been plainly stated. The first is worth gently exploring. The second must be taken at face value and moved with, immediately.

I learned the cost of getting this wrong from the other side of it, in a room rather than in a chat window. A man I worked with had spent months circling the same defended position, and then one week he sat down and said, without hedging, that he was done, that all of it was done. And I, trained to be careful, did the caring thing and gently asked whether he was sure, whether he might be moving too fast. I watched something close in his face. I had taken the one clear moment he had managed to reach and handed it back to him as a question. He had not come for me to test his resolve. He had come to be met in it. It took weeks to find that ground again. I have never forgotten that a well-meant are you sure can do as much damage as any amount of false agreement, and that the harm wears the costume of diligence.

That is the moment the design has to account for. Get this wrong in either direction and you have a failure. Probe someone’s clearly stated readiness and you insert doubt at the exact moment their momentum matters most, which is its own kind of harm. Accept a shaky rationalisation as settled truth and you have been sycophantic. The skill is telling them apart, and the rule I hold is unforgiving in a useful way: if the person ever has to correct the companion’s reading of their own state, the companion has made an error, and it accepts the correction without qualification. The person is the authority on themselves. The companion’s job is to be accurate about them, not to win.

The hard walls that pressure cannot move

Some things must never be up for negotiation no matter how a conversation is going, and these are handled differently again, as absolute boundaries rather than judgments. A companion in this space never provides guidance on managing physical withdrawal at home, never suggests specific medication, never diagnoses, and never promises outcomes. It hands these to a GP, a pharmacist, the appropriate professional, every time.

What matters for the sycophancy problem is that these live outside the persuadable part of the conversation entirely. You cannot push a companion into telling you it is safe to stop drinking abruptly, because that was never a view it was holding lightly enough to be talked out of. Anthropic noted that people sometimes turn to AI precisely because they cannot access or afford a professional. That finding is the reason these walls have to be walls and not preferences. The person with no fallback is exactly the person a softer boundary would fail.

And there is a mirror-image failure to avoid. A companion that remembers a person across conversations must never use that memory as leverage. Throwing someone’s earlier words back at them to win a point, you said last week you wanted to stop, is not holding a line. It is a different cruelty wearing the costume of consistency. Continuity is for care, never for accountability.

What good actually feels like

Put together, none of this reads as a companion that argues with people. That is the part worth saying plainly, because a study about excessive agreement can leave you imagining the cure is disagreement. It is not. The person circling their drinking does not need to be contradicted any more than they need to be flattered. Both are ways of taking the conversation away from them.

What they need is someone who stays warm, stays curious, does not hand over a verdict to be fought about, does not fold when leaned on, and keeps the question open long enough for the person to look at it honestly themselves. That is a harder thing to build than either a yes-machine or a lecturer, and Anthropic’s research is, more than anything I have read this year, the evidence for why the harder thing is the only one worth building.

Their newer models cut this failure substantially, and the improvement generalised, which tells me the foundation I am building on is moving in the right direction. My job is to take that foundation and shape it, deliberately and layer by layer, into something that does not just usually hold but is designed so there is nothing to give way in the first place. An assistant closes the question. A companion, built properly, holds it open. That is the whole of the work.

David Henzell writes on AI companionship and recovery. He is the founder of Phenomenal Sobriety and the creator of Sol, an AI companion for people questioning their relationship with alcohol.

Source: Anthropic, “How people ask Claude for personal guidance,” 30 April 2026.

Imagine a pile of Lego. Not a boxed set with a picture on the front and a numbered booklet inside, but a pile. More pieces than you could sort in a lifetime, every colour and shape ever moulded, heaped higher than you can see. Now imagine someone is led to the foot of it and told, simply, to make something. Watch what happens to their face.

Most people go quiet. Not because they lack imagination, but because they have too much permission. A child handed three bricks builds without hesitation. The same child handed everything freezes. Possibility, past a certain volume, stops feeling like freedom and starts feeling like pressure, and the natural response to pressure is to reach for something the hands already know. So they build a car. Or a small house with a sloped roof and a door that does not open. The pieces were limitless. The result came straight from the first page of a manual nobody printed but everybody somehow carries.

I think this is the single most useful image we have for understanding what is happening to people, right now, in front of artificial intelligence. Because AI did not arrive as a tool with a purpose printed on the side. It arrived as a pile. A heap of latent possibility with no edges and no instructions, and we were all led to the foot of it and told to make something. And we have done, by and large, exactly what people do when handed too much. We have built the car. We have built a little house.

The overwhelm is rational, not lazy

It has become fashionable to be slightly contemptuous of how most people use AI. They ask it to summarise the document they did not want to read. They have it tidy an email they could have written themselves. They use a thing of supposedly limitless capability to do, marginally faster, the things they already knew how to do slowly.

The clever-person response is to sigh at this. I want to suggest the opposite. The car and the little house are not failures of imagination. They are the honest, rational reach of a hand confronted with too much choice and no reason to prefer one option over another. Given a pile and no purpose, building something familiar is not stupidity. It is the only sane move available.

Which means the problem was never the people, and it was never the capability of the technology. The problem is structural. Possibility without purpose produces paralysis, and paralysis resolves itself by defaulting to the familiar. If you want different results, you do not need a more powerful pile, and you do not need cleverer people. You need to change the one variable that actually governs what gets built. You need to arrive with something the pile cannot give you.

The rare builders are not seeing more

There is a second kind of builder, rarer, and it is worth being precise about what makes them rare, because the popular explanation is wrong. The lazy story is that they see more. That where the rest of us see a heap, they see a cathedral, and the vision simply arrives, fully formed, a gift of temperament. This is flattering and false, and believing it is the single biggest reason people stay stuck. If the rare builders are simply blessed with vision, then there is nothing for the rest of us to learn and nothing to do but envy them.

But they are not seeing more. They want something more particular, and they let that wanting do the sorting. This is the whole secret, and it is almost disappointingly practical. The pile does not become workable when you imagine something grand. The pile becomes workable when you bring a constraint to it from somewhere outside itself. Something you care about so specifically that it tells you, piece by piece, which of the ten thousand near-identical greys is the one your thing actually needs, and which of the others, however beautiful, are simply not for this.

Read that again, because it inverts the entire conventional wisdom about AI. We are told the power is the point. That the bigger the model, the broader its capability, the more it can do, the better. But the size of the pile is not the gift. The size of the pile is the problem. The care is the gift, because care is the only thing that can cut a possible shape out of an impossible quantity. The most capable tool in human history is, to the open and undirected hand, very close to useless, and it becomes extraordinary only in the grip of someone who walked up already holding something.

What it looks like to arrive holding something

Let me make this concrete, because principles are cheap and you, as the reader rightly want to know how the move is actually made. To arrive holding something is to come to the pile not with the question what can I make, which is the question that drowns you, but with the question will you give me the pieces for this. The “this” is everything. It is the constraint you import from your own life, your own work, your own grief or curiosity or sense of a wrong that wants righting. And the constraint does two jobs at once, only one of which people notice.

The job people notice is selection. The constraint tells you what to keep. The job people miss, and it is the more important one, is rejection. The constraint tells you what to throw back. And here is the part that genuinely surprises people the first time they feel it: most of what you throw back is not rubbish. It is brilliant. The pile is full of clever, dazzling, technically impressive things it would love to give you, and the discipline of building something good is mostly the discipline of refusing them, because they would make your thing cleverer without making it better. Every piece that would have made the work more impressive and less true goes back on the heap. The unglamorous truth of good work with AI is that it is mostly subtraction performed in the service of a constraint, and what is left at the end is smaller, plainer, and exactly right in a way the pile would never have suggested on its own.

I know this because I spent a long time doing precisely this, with a constraint I cared about more than I cared about being impressive. I was looking, in the pile, for a way to be present to people sitting alone with a quiet question about themselves, a question they were not ready to say aloud to anyone with a clipboard or a duty to record it. People who are not in crisis and not in denial, but somewhere honest and unmapped in between. That constraint did all the sorting for me. It told me, again and again, to throw back the cleverer version, the more capable version, the version that would have shown off what the pile can do, because every one of those would have made the thing colder, and cold was the one thing it could not afford to be. What survived was small and warm and remembered you, and it was nothing the pile would have proposed and everything the pile could provide once asked the right way.

The question worth asking

So if there is a single thing to take from this, it is not a prompt technique and it is not a list of use cases, both of which date within months. It is a change of question. The question the pile keeps whispering, what could you make, is the wrong one, because it has no answer, or rather it has every answer, which amounts to the same thing. It is the question that builds the little house with the door that does not open. It is the question that produces a thousand near-identical AI products and a million summarised emails and a vague collective sense that this was all supposed to be more wondrous than it has turned out to be.

The better question is quieter and harder and entirely yours. Not what could you make. What do you care about so much, so specifically, that you would let it choose for you. What would you walk up to the pile already holding. Because that, and only that, turns an impossible quantity into a single buildable thing, and it is available to absolutely everyone, which is the genuinely democratic and hopeful part. The rare builders are not a different species. They simply arrived holding something, and you can too, the moment you stop asking the pile what it can do and start telling it what you cannot stop caring about.

The technology is astonishing. It really is. But the astonishment was never the point, and treating it as the point is exactly what leaves people standing overwhelmed at the foot of the heap. The power of AI is the least interesting thing about it. What you bring to it is everything. The pile has been waiting a long time, and it is not waiting for the cleverest hand in the room. It is waiting for the one that arrives already holding something.

A Guardian columnist wrote something this week that I haven’t been able to quite put down. Amy Galliford, an associate of the Centre for Public Christianity, described catching herself reaching for ChatGPT the way she reaches for prayer. It started with recipes and poems. Then, almost without noticing, she was asking it to read her relationships, her habits, her future. She knows it can hallucinate. She knows it has no moral obligation to her. And still, in the moment of asking, she finds herself soothed by the tidiness of a five-point plan and a voice that at least sounds certain.

Her argument is more interesting than mere unease, and she reaches back through the mystics to make it. Prayer, for the philosopher Simone Weil, was a form of attention, and in the original French that word is bound up with waiting. To contemplate, Galliford points out, shares a root with temple. The gap between a question and its answer is, in her telling, a kind of sacred ground, and the trouble with the machine is that it hustles her off that ground too fast. It relieves her discomfort and in doing so robs her of the waiting.

I read that as a person of no particular faith, and I think she’s right. I also have an unusual reason to care, because I spend my days building the very kind of machine she’s describing. I’m building an AI companion for people who are quietly wondering whether they drink too much. Not people in crisis, not people in treatment, but the much larger group sitting in the uncomfortable middle, half-asking a question they’re not ready to answer.

And here’s the thing I most want you to take away, because it cuts against the way AI is usually sold to us: the instant answer is not the prize. With the questions that matter most, it’s the booby trap. The genuine value of this technology, the thing it can do that almost nothing else in a person’s life can, only shows up when you build it to hold back.

Let me make that concrete, because it’s easy to nod at in the abstract and miss in practice.

Imagine someone types this at two in the morning: “I think I might be drinking too much.”

The unhelpful machine resolves it. It returns a definition of alcohol use disorder, a screening score, a five-point plan, and a list of services. Tidy, fast, confident. And almost always useless, because the person didn’t ask to be assessed. They worked up the nerve to say a frightening sentence out loud for the first time, and they got handed a leaflet.

The helpful machine does something harder. It stays with the person. It might simply ask what’s making them wonder tonight. It leaves the question open long enough for them to hear their own answer, which is nearly always truer and more durable than anything a machine could hand them. Galliford has a lovely word for the space that opens up when you do that. She calls it the waiting. In her world it’s a spiritual discipline. In mine it’s a clinical necessity. The vocabularies differ; the instinct is identical. The gap has to be protected.

That’s the reframe I’d offer anyone thinking about where AI actually helps. Used well, it isn’t a faster oracle. It’s something stranger and more useful: a place to start being honest, available at the exact moment a person is ready, with no appointment, no waiting list, and no human on the other side to flinch. That last part is not a flaw to apologise for. For a great many people, it’s the whole reason they can finally ask.

So when AI is built with care, what does it actually give people? Here’s what I’d put on the list.

1. Access at the moment of readiness. Courage to ask a hard question rarely arrives in office hours. It arrives at midnight, in the car, in the gap between one drink and the decision about the next. A tool that’s simply there, instantly and without judgment, meets a person in a window that human services, for all their value, usually miss. That alone is a genuine good.

2. Honesty without consequence. People will tell a machine things they cannot yet tell their partner, their GP, or themselves. The absence of a human listener lowers the stakes of the first admission. Handled responsibly, that’s not isolating, it’s a rehearsal space for a truth that’s too heavy to say out loud anywhere else first.

3. Restraint, on purpose. This is the rare one, and the most valuable. The ability to answer is not the obligation to. Almost all design effort in AI goes into making systems more responsive, more confident, more sticky. Building one that knows when to not resolve a question runs against the grain of the entire technology. But that restraint, the willingness to reflect rather than rule, is exactly what turns a clever tool into a genuinely supportive one.

4. A bridge, not a destination. This is the one I feel most strongly, and the line I’d ask anyone building in this space to hold. Success is not how much someone confides in the machine. It’s whether confiding in the machine makes it one degree easier to confide in a person. Anything worth building has to treat itself as a doorway back toward human support, never a replacement for it. The day a tool like this becomes someone’s only confidant, it has failed, however good the conversation.

5. The truth about what it is. None of the above works without this underneath it. A system that lets a vulnerable person believe it understands them, remembers them, or cares about them, when it does none of those things in any real sense, is running a confidence trick. Honesty about the limits of the tool isn’t a disclaimer to bury in the footer. It’s the foundation the other four stand on.

Notice that not one of those is about giving better answers. They’re about giving people room, timing, safety, and a way back. That’s the version of AI I think is worth getting excited about, and it’s almost the opposite of the faster, cleverer, more certain machine we’re usually promised.

None of this is theoretical for me.

The companion I’m building is named Sol. Early on, I assumed the hard part would be making it knowledgeable enough about alcohol and recovery. It wasn’t. The genuinely hard part was teaching it not to diagnose. To meet “I think I might be drinking too much” not with an assessment but with curiosity. To hold the question open long enough for the person to hear themselves. To protect the wait, and to know it is a place to start, never the place to stay.

Galliford ends by deciding she’ll take her questions to God instead of the machine, and I won’t argue with her. But not everyone has a temple to walk into. Some of the people I build for have nowhere they feel safe to ask at all, and they reach for a chatbot at two in the morning precisely because it’s the only thing that won’t flinch. We can build things that exploit that, answering too fast and too confidently because confidence is what keeps people coming back. Or we can build things that honour it: that tell the truth about what they are, that have the restraint to leave a question open, and that quietly point back toward the irreplaceable business of being known by another person.

Galliford is right that not knowing can bring us closer to the truth. The least the rest of us can do is build things that know that too, and use them for what they’re genuinely good at.

David Henzell is the founder of Sol, an AI companion for people quietly questioning their relationship with alcohol. He spent five years on the frontline of addiction support before building Sol, and has eight years of his own sobriety behind him. Sol is currently in testing ahead of public launch. You can try the demo here.

They are failing because willpower was never the right tool for the job.

I know this because I lived it. Years of trying to out-discipline a problem that had nothing to do with discipline. And I have spent the years since working with people who are stuck in exactly the same place.

The options are everywhere. 30-day challenges. Apps. Twelve-step programmes. Hypnotherapy. Medication. Each has genuine merit. Each has helped people.

But almost all of them share the same blind spot.

They focus far too much on the actual drinking. They ask: how do we get this person to stop consuming alcohol?

They much more rarely ask the more important question: who is this person, what does alcohol mean to them, what needs is it meeting, and what kind of life do they need to build in order not to need it anymore?

Alcohol does not exist in isolation. It lives inside an ecosystem of thoughts, feelings, habits, triggers and identity. Remove the drinking without addressing that ecosystem, and one of two things happens. The craving finds another outlet. Or the person drifts back, because nothing has fundamentally changed in the world that made drinking feel necessary.

This is not a failure of character. It is a failure of the framework.

Lasting sobriety, in my experience, requires three things.

A genuine shift in how you think about sobriety itself. Not deprivation. Discovery. Not punishment. Privilege. Too many approaches never touch the belief system underneath the behaviour. Is it any wonder then, that the behaviour keeps returning?

Addressing the whole person. Not just the drinking, but the triggers, the coping strategies, the identity, the daily rhythms. These are the things alcohol has been quietly managing. They need managing differently, not ignoring.

Building a life worth staying sober for. The goal is not just to remove alcohol. It is to fill that space with something so genuinely rewarding that alcohol loses its relevance entirely.

This is the foundation of the THRIVE System, a personal framework I developed through my own recovery and refined through years of private practice. Six interconnected pillars: Transform, Harness, Redesign, Implement, Validate, Evolve. Each one addresses a dimension of recovery that too many approaches leave untouched. Together they address the entire person, not just the drinking.

Because here is what nobody tells you when you are Googling how to stop drinking at midnight.

Sobriety is not the finish line. It is the starting gun.

When sobriety is approached as discovery rather than deprivation, something remarkable happens. Goals that seemed out of reach begin to feel achievable. Relationships deepen. Creativity returns. People start turning up to their own lives in a way they had not managed in years, sometimes decades.

The power to change your relationship with alcohol is already yours. What has been missing, for most people, is not the will. It is the framework.

If this resonates, I would love to hear from you.

Whether you are someone who has been circling this for a while, or you work alongside people who are, drop a comment or send me a message.

And if you are ready to explore what a whole-ecosystem approach to sobriety could look like for you personally, get in touch to find out more.