Lumi - ADHD AI Companion
A new day for your brain
Six months ago I watched someone I care about spiral into burnout — not because they weren't trying, but because every system they used was built for a neurotypical brain.
They were smart. Self-aware. And completely frozen.
Therapy waitlist: 4 months. ADHD coach: $350/month. Generic AI: technically works but feels like asking a spreadsheet for emotional support.
I'm a developer. I built the thing I wished existed.
The problem I kept coming back to
Adults with ADHD don't lack intelligence or motivation. They lack executive function scaffolding — the external structure that helps them start, stay regulated, and re-enter when they fall off.
Therapy gives you 1 hour a week. That leaves 167 hours unsupported.
Every productivity app I looked at made the problem worse. They require consistent habits to maintain — which is exactly what ADHD makes impossible. Notion, Todoist, even the ADHD-specific ones — they all assume you'll show up consistently. ADHD doesn't work that way.
So I built Lumi. An AI companion that meets you where you are each day. No setup. No template. Just open it and talk.
What Lumi actually does
Daily mood-first check-ins (how you feel before what you need to do)
Task paralysis support — breaks the freeze with one next step, not a 12-item list
AI body-doubling for focus sessions
Voice brain dump — speak your thoughts, Lumi organizes them
Pattern memory — tracks your energy, mood, and focus windows over time
Weekly Brain Report — surfaces your patterns so you can understand your brain, not fight it
The AI is built on Claude, trained specifically for neurodivergent patterns, with a defined persona. Not a raw chatbot. A companion.
What I learned building in this space
1. Shame is a product problem.
Every friction point in a mental health app is amplified by shame. If it's hard to re-enter after missing a day, users don't come back. We built explicit "welcome back, no guilt" flows. Zero streaks. Zero punishment mechanics.
2. The comparison isn't other apps — it's therapy.
Early on I was pricing against Headspace and Calm. Wrong frame. The user isn't choosing between apps. They're choosing between Lumi at $14/month and an ADHD coach at $350/month. Price and position accordingly.
3. Voice input changed everything.
Users who could brain dump by talking retained dramatically better in testing. Typing requires executive function. Speaking doesn't. This one insight changed the core UX.
4. Re-entry matters as much as onboarding.
ADHD users will hyperfocus on your app for a week and then disappear for two weeks. Most apps treat that as churn. We treat it as normal. The re-entry experience — coming back after being gone — is as important as the first-time experience.
5. "No credit card required" isn't optional.
ADHD brains have decision fatigue and trust issues with subscriptions after years of forgetting to cancel things. Remove every barrier to starting the trial.
Where we are
Just submitted to the App Store — waiting on Apple review
7-day free trial, no credit card required
Core: $14/month | Companion: $24/month
Early users from TestFlight have been the most validating feedback of my career. One message I got last week:
"This is the first thing that's ever talked to me like someone who actually gets it."
That's why this exists.
What's next
iOS App Store launch (pending review)
Content strategy around ADHD support pages (task paralysis, RSD, burnout, executive function)
Building the weekly email report into a proper retention driver
Happy to talk shop on building in the mental health / neurodivergent space, AI companion product design, or anything else. This community has been a huge resource — giving back where I can.
Try it free: https://lumimind.app
6 Comments
This is one of the sharpest build-in-public posts I have read, and the deepest insight in it is the one you list fourth: re-entry matters as much as onboarding. That is not just a UX lesson, it is your entire positioning, and nobody else can take it from you. Every habit app optimizes the first impression and then punishes the return with a broken streak. You built the one app that is glad you came back. Fuse your own pieces, the enemy (tools that demand the consistency ADHD removes), the no-shame re-entry, the zero streaks, into one line: "the companion for the brain that does not show up on schedule, and never makes you feel bad about it." That is a promise Notion and Headspace structurally cannot make.
Your testimonial also tells you what the product actually is, and it is not the six features. "The first thing that has ever talked to me like someone who actually gets it" means the product is the relationship, and the features are only how you deliver it. Which means your moat is not the persona alone, because a competitor can ship a warm ADHD tone next quarter. It is the pattern memory: after a month, Lumi knows your energy windows, your triggers, your Tuesday spiral, and a cold competitor cannot. Lean into the thing that compounds, that it does not just get you, it gets you more every week. That is a switching cost no fresh app can cold-start.
You already nailed the pricing frame, therapy and a coach rather than Headspace, so keep it. One thing to hold as you build content around RSD and burnout: stay unmistakably the companion between sessions, not the clinician, because that is both the safer lane and the sharper one. And one question worth answering as you launch: do your best early users stay for the daily support, or for the weekly report that shows them their own patterns? Because whichever one creates the "it gets me" moment is the thing to build the entire retention loop around.
The point about re-entry being just as important as onboarding really stood out to me.
A lot of apps treat someone disappearing for a week as churn, when in reality life just gets messy. Especially for users dealing with burnout, overwhelm, or executive dysfunction.
It feels like the real product challenge isn't helping someone on their best day, it's designing for their hardest day.
I'm curious, what did your early testers respond to most: the voice brain dumps, the weekly reports, or simply having a companion that welcomed them back without guilt?
This is one of the stronger AI companion use cases because the product is not just “chat with an AI.” The real value is the re-entry loop: someone disappears, comes back with shame or overwhelm, and the product has to meet them without making them feel like they failed.
That is also where the naming matters more than usual. Lumi is warm and simple, but it may become harder to own as the product expands into ADHD support pages, weekly brain reports, emotional regulation, voice brain dumps, and companion-style retention. It feels friendly, but the broader category you are moving toward is closer to a trusted support layer for neurodivergent adults.
If this is still before App Store launch and public content starts compounding, I would pressure-test whether the product needs a more ownable brand shell now, before users, SEO pages, emails, and App Store memory lock in around Lumi.
Lyriso .com would fit this direction well because it has a softer care/trust feel without boxing the product into productivity, therapy, or generic chatbot territory. It could carry mood check-ins, ADHD support, emotional scaffolding, reports, and companion features under one calmer brand.
Hey Brandon,
Spent ~10–15 min on Lumi's site today. The copy “Starting it is the hard part” is the kind of line only someone who’s lived this would write. The brand voice throughout feels very real, most AI apps borrow empathy, yours actually feels lived.
One small thing I’d test, since it’s mobile-first: I didn’t see App Store / Google Play badges above the fold. That moment — when someone decides “I want this” is where even a second of friction can lead to bounce, especially for ADHD users.
Feels like a simple CTA placement experiment with real upside.
Happy to share 3–4 more conversion-focused observations if useful.
— Jitesh (designer @ Microsoft, currently obsessed with AI-product UX)
Thanks Jitesh — the lived-in copy is the whole bet so it means a lot that it lands. You're right about above-the-fold friction; we had a surprise issue during a final review cycle with Apple, which threw us off track for a few days. In the meantime, i'm working on a few site updates as I type this, including removing the badges for a pre-launch email form). For now I think a waitlist CTA above the fold makes more sense than a dead App Store badge — but once approved, badges go up the same day.
Would love to hear those 3-4 observations.
Cheers!
Brandon Holt
Brandon,
First off, that Apple review timing is brutal. Sorry it threw you off for a few days. And agreed, a waitlist CTA above the fold beats a dead App Store badge every time, so good call making that swap while you wait on them.
Since you're already in there making changes, here are a few things I'd keep in mind for the email version, all at the decide moment and tuned to your ADHD users.
1. Watch the microcopy under the signup. If it leads with something like "you agree to receive marketing emails," that consent line first adds a beat of hesitation right when someone's deciding. I'd lead with the win instead, something like "Be first in. We'll email you the day it's live. No spam," and tuck the legal smaller underneath. Same info, less friction.
2. Whatever the new waitlist button becomes, give it your voice instead of a default "Join the waitlist." Something that sounds like Lumi, maybe "Save my new day" or "Get the first day." Small thing, but the voice is your whole edge, so I'd carry it into the most important click. (Same goes for the "Try Lumi Free For 7-Days" buttons lower down. Repeating them on a long page is right, I'd just vary the wording so it stays fresh instead of the same flat line every time.)
3. Your proof is genuinely strong, the numbers, the testimonials, the press. The only thing I'd eyeball is whether any of it sits right next to the signup, or only further down. Proof at the decide moment converts harder than proof you scroll to.
If I had to pick one to start with, it's #1. Cheap to change, and it pulls a little hesitation out of the exact moment people commit.
Side note, since you asked for more. The reason I spot this stuff fast is I've been building a tool that does exactly this, reviews the interaction details on a live app and hands you the actual fix code. Real, growing apps like Lumi are exactly who I built it for. Happy to run it on Lumi and send what it catches, totally up to you.
Rooting for the Apple approval. Tell me when it's live and I'll be first on the badge.
Best,
Jitesh