Products · Olive
Olive: Building a Bridge to Real Care
A mental health companion my brother and I built around one question: when someone is ready for more than a conversation, how do you help them take the next step toward real support?
Muhammad Gane12 min read
My brother Jameel had the name first. He called the idea Olive, drafted an early picture of what it might look like, and knew what he wanted it to do: give someone a place to talk through what they're carrying, and help them move toward real support when they're ready for it.
Neither of us came to the subject in the abstract. We've seen these struggles close to home, and that made mental health personal for both of us long before Olive had a name. Jameel also works in the mental health space through the board roles he holds. By the time he described the idea to me, it didn't feel like a side project. It felt necessary, and larger than either of us, and we both wanted to see it through.
What he needed was someone to build it. By then I was writing code nearly every day, and his invitation was about as direct as brothers get: you're already coding every day, so bring this to life. That's my memory of it rather than a transcript, but the substance hasn't changed in the retelling. I said yes because I already cared about what it was for.
Olive is an AI mental health companion for adults. It offers supportive conversation, remembers what you allow it to, and, when you ask, helps you find professional care or community support and prepare to reach out yourself. It isn't therapy, a diagnosis, or an emergency service. Jameel is Olive's co-founder and CEO. I'm its technical co-founder, and I've written all of its software, but Olive has been a joint effort from the start: his direction, his feedback on almost every part of it, and the nights he spent working alongside me shaped what it became. It's the largest thing I've shipped, and it means a great deal to me. This is the story of how we built it, told from my side of the work.
The product
Olive
Support in the moment. A bridge when you need more.
A mental health companion for adults, free to download on iPhone and Android during its Canada-first beta.
The step after talking
When we talked about what Olive should be, we kept returning to one particular gap. Plenty of people can talk about what's weighing on them: to a friend, in a journal, sometimes to a chat window. Fewer find it easy to turn that into a next step.
The gap isn't one thing. It can be not knowing what kind of help to look for, whether that's a counsellor, a psychologist, a peer group, or someone who works with grief. It can be the prospect of explaining everything again, from the beginning, to a stranger. Sometimes it's as small and as stubborn as not knowing what to write in a first email.
For a lot of people, the hardest part comes earlier, and it comes from the people around them. We've seen it in our own lives and communities. In many families, cultures and communities, struggling with your mental health gets treated as weakness, or as something that isn't real. Young men hear a familiar version of it: you're a man, suck it up. Depression and anxiety get talked about as one person's failing, when they can reach anyone. When that's the message from the people closest to you, even naming what you're carrying can feel like a risk, and contacting a professional can feel out of the question.
Those were the problems we chose to work on, and I want to be careful about how much weight they can carry. A lot of what stands between a person and care has nothing to do with an app. An app can't change what a family believes about mental health. Cost matters. So do waitlists, whether anyone nearby does the right kind of work, and whether it fits someone's life right now. Software can't create a counsellor or cover a fee. What it can do is make the reachable part a little easier: knowing what to look for, having your own words ready, and not starting from nothing.
A conversation seemed like the right way in because it asks so little up front. You don't have to know what you need before you begin, or justify it to anyone first. You can say something half-formed and come back to it later. We never wanted Olive to become the relationship in someone's life, or to stand in for the people already in it. We wanted a low-pressure place to start, one that points outward.
What the purpose asked of the product
Once the goal was clear, most design decisions came down to a single question. Does this help someone take the step they want to take, or does it only keep them talking? A few answers to that question ended up shaping almost everything else.
Listening before a plan
The easiest thing for a conversational product to do is be helpful immediately. Someone mentions they're exhausted, and a sentence later they have five tips about sleep. It sounds supportive. Often it misses what the person wanted, which may have been to say the thing out loud and have it land.
So we designed Olive's conversations to listen first. Early in a conversation it tries to understand; later, when the moment seems right, it might offer one concrete idea. Not every response will strike that balance, and we don't expect any system to manage it every time. But the intention sets the pacing, and pacing is a large part of whether someone feels heard or handled.
Finding Olive's voice was the most collaborative part of the build, and it's where Jameel's hand shows most. I'd write a first version of who Olive should be: how it speaks, what it asks, when it holds back. Jameel would take it apart and rework it, often substantially. I'd build his changes back in, test them against the standards we'd set for safety and tone, and hand it back to him. We went around that loop many times before Olive sounded like itself.
People also choose how Olive talks with them. There are three conversation styles: Olive, which is warm and helps you think things through; Socratic, which asks gentle questions one at a time; and Values, which connects a small next step to what matters to you. They're ways of having a conversation, not clinical programs. Preferences such as a nickname, tone, focus areas and topics you'd rather leave alone work the same way. They're things you tell Olive, rather than things it guesses about you.
Continuity you can see
Repeating yourself is tiring, especially about something hard. Continuity across conversations is one of the most useful things a companion can offer, and one of the easiest to get wrong. An app that remembers the context you've shared is very different from one that quietly builds a profile you can't inspect.
Memory Vault is how we tried to keep that difference visible. What Olive remembers is shown to you, organized by category. You can change it, pin what matters, or remove something entirely. You can turn off a whole category, or switch memory off altogether.
- See it
- What Olive remembers is listed, grouped by category.
- Change it
- Correct a detail, or pin the ones that matter most.
- Remove it
- Hide or forget a single memory.
- Switch it off
- Turn off a category, or memory as a whole.
Those controls describe what the feature does, and we've tried not to let them imply more. Switching memory off stops Olive carrying context from one conversation into the next. It doesn't mean nothing is stored or processed in order to run the service, and the privacy policy explains that side plainly. I'd rather someone understand the real boundary than feel reassured by a vague one.
Different kinds of help
"I think I should talk to someone" can mean very different things. A counsellor and a peer-support group are both real help, but they're different next steps, found in different places, suited to different moments. A list that mixes them looks generous and makes the choice harder.
Olive keeps them apart. Find Care is for professional support: therapists, counsellors, psychologists, clinics. Find Support is for community and peer resources, such as support groups, warm lines, and grief or recovery support. If a request could mean either, Olive asks which one you mean before it searches.
It also searches when you ask it to, instead of deciding on its own that you sound like you need a referral. Being in a hard conversation isn't the same as asking to be sent somewhere, and the person should be the one to start that search. Neither search can promise the right result. The resources are Canada-first, availability changes, and some places simply have less. Where hours or eligibility are uncertain, the aim is for Olive to say so rather than paper over the gap.
Canada is where we started, though, not where we want this to stop. Stigma and the distance to care aren't Canadian problems, and over time we'd like the parts of Olive that help people find and reach support to work for people in other countries too. That's an ambition for later versions, not something Olive does today.
A conversation that can leave with you
The last decision grew out of the others. If a conversation is where someone starts, it shouldn't be where everything they worked out stays. What they want help with, and how they'd put it in their own words, is worth carrying to the person they eventually contact. That idea became Care Bridge.
Care Bridge
Care Bridge is the part of Olive that most directly expresses what Jameel first described to me, so it's the one worth walking through.
Say someone has been talking with Olive about work. It's been relentless for a couple of months, their sleep is slipping, and through Find Care they've found a counsellor they'd like to contact, but they aren't sure how to begin. With Care Bridge, they can turn the relevant parts of that conversation into a short brief: what they want help with, what they're comfortable including, and a few sentences they might use when they call or write to someone for the first time.
From the conversation
Things you said, in your words
- Work has been relentless for a couple of months.
- I'm not sleeping well.
- I'd like to talk to a counsellor, but I don't know how to start.
A note you can edit
You choose what to include
Brief for a counsellor
Draft · every line editable
- What I'd like help with
- Managing work stress that has been building for about two months, and getting my sleep back on track.
- What I'm comfortable sharing
- It's mostly work. I haven't spoken to anyone about this before.
- What I might say
- “Hi, I'm looking for a counsellor who works with stress and burnout. Are you taking new clients, and what does a first session cost?”
A next step you choose
You decide whether to share
Copy it, share it, bring it to a first call, or keep it for later.
Olive doesn't send it, book anything, or contact anyone for you.
The brief is theirs. They can rewrite any of it, take things out, or set it aside for another day. Nothing is shared on their behalf. It goes to someone only if they choose to copy it, share it, or bring it into a first conversation themselves. Olive doesn't book an appointment, contact a provider, or establish a care relationship for anyone. The first call still belongs to the person making it. What we can do is make that call feel a little less like starting from zero.
That choice is about who sees the brief, and it's separate from the processing any AI conversation needs. To generate a reply or draft a brief, the conversation is sent to Olive's backend and to the AI providers that power it, and Olive asks for consent before any of that begins. The privacy policy covers the details, including the limits.
Care Bridge is the clearest answer we have to the question we started with. The conversation matters. The brief is the part that can travel.
Building it
I spent roughly seven months building toward Olive's first release. The idea is older than the code; Jameel had been shaping it before I wrote a line, and he stayed close to the work the whole way through. My job was to turn what we'd imagined into something a person could actually use, which in practice meant one mobile app for iPhone and Android and the backend it relies on.
The first version reached the App Store at the end of May 2026. The first update followed a day later, and it was about unglamorous things: getting started with the app, and how a conversation felt to use. That set the pattern for most of what came after.
A lot of the work since has been at that scale. Is it clear when Olive is still composing a reply? Is a saved note easy to read a week later? Can you tell an empty Memory Vault apart from one that hasn't loaded? Can a step be removed entirely? None of these would make a feature list, and in a general-purpose app some of them could reasonably wait. In something people open when they're already having a difficult time, an unclear loading state or an unnecessary screen costs more than it seems to.
Later work has focused on continuity, on making memory easier to understand, and on making the path toward human support more deliberate. For the next release, I've been building My Plan, where a person writes down their own practical steps for a hard moment: how they notice they need support, what helps them settle, and who they can reach. I've also been working on a more considered handoff to crisis support. Until they ship, please treat both as work in progress.
On crisis, our position has been simple from the start. Olive isn't a crisis service and shouldn't try to behave like one. If someone may be in crisis, the priority is putting a real human resource in front of them, which in Canada means 9-8-8, not keeping them in a conversation with an app. Olive doesn't place a call or send a text on anyone's behalf. It's an area I test carefully, and one where we'd much rather understate than overstate what software can do.
Throughout the build we held Olive to a bar for safety and tone, and I checked changes against it. That discipline matters. It also isn't clinical validation, and it doesn't make Olive a clinician. Both statements are true, and the second doesn't cancel the first.
What building it taught me
Memory is only useful if you can manage it
It's tempting to think of memory as a capability. Building Olive, I came to see it as a responsibility with an interface attached. Remembering helps only when a person can tell what's remembered and how to change it. Otherwise the same feature that spares someone from repeating themselves becomes a reason to trust the app less. That's why the controls get as much design attention as the remembering itself.
Finding help and reaching it are different jobs
A search result gives you an address. It doesn't give you a first sentence. Much of the difficulty sits in the space between having a name and actually getting in touch, and a better list doesn't close it. Care Bridge exists because that space deserved its own tool, one built around the person's own words.
Small problems aren't small here
When the underlying AI works, it's easy to treat everything around it as polish. I don't think that holds for this kind of product. A slow reply with no sign of progress, a note that's hard to read, one confirmation screen too many: each asks a little more of someone who may not have much to spare. Some of my most worthwhile work on Olive has gone into things nobody would ever list as a feature.
Deciding where the software stops
The scope decisions I care most about are the ones about what Olive doesn't do. It doesn't diagnose, treat, or call itself therapy. It doesn't search for help unless asked, send anything on someone's behalf, or contact a provider. It doesn't use streaks or guilt to bring people back, and it doesn't try to stand in for a crisis line. Some of those are safety decisions and some are about respect, but they share a principle: the person stays in charge of their own next step. Holding that line has made the product clearer to build and, I hope, easier to trust.
Where it goes from here
Jameel's original idea was a companion that helps someone move toward real support. That's still the test we apply to every change. Most of what we're working on now points the same way: helping a person carry their own context and practical steps forward, with more control over what's kept and less friction on the way to reaching out.
We're still early. Olive is a Canada-first beta, and there's plenty we're still trying to get right. But it's real, people can download it, and it grew out of something we both cared about long before it had a name.
If you'd like to see what we've made, Olive is free to download on iPhone and Android.