Why I built an AI Coach into my expense tracker instead of just another dashboard
I've been building SmartSpendr, an expense tracker with an AI Budget Coach baked into the core of it.
The honest starting point: I wanted tracking expenses to feel less like a chore. Most trackers give you a list of transactions and a chart, and that's it. You're on your own to figure out what to actually do with that info. I wanted AI to close that gap, so instead of just showing data, the Coach reads your spending by category, wallet, and time period and gives you plain-English direction on what to change.
It's live now on iOS and Android — free to start, with the AI Coach, CSV/XLSX import, and multi-currency tools in the $10/mo or $100/yr Pro tier.
Where I'm actually at: early days, small number of installs, no reviews yet. Posting here to start documenting the build in public — what's working, what isn't, and what I'm learning as I go.
Would love to hear from anyone who's built AI features into a consumer app — especially around getting people to actually trust and act on AI-generated advice rather than just glance past it.
The trust question feels especially important here because the Coach isn't just explaining someone's spending — it's asking them to change behavior based on its judgment.
Getting someone to read an AI insight and getting them to act on it seem like two very different thresholds.
Curious which one you're seeing break first once people actually use it.
Really good distinction, and honestly I don't have clean data on this yet — small sample so far. My gut says it's the second one that breaks first. Reading an insight and nodding along is easy because it's about the past ("you spent X on Y"). Acting on it means trusting the Coach's judgment about the future, which is a different kind of ask.
The thing I'm trying to get right is making the first insight someone acts on low-stakes and obviously correct — something like "you have 3 subscriptions you haven't used in 60 days" rather than "you should cut your dining budget by 20%." One is a fact wrapped in a suggestion, the other is a judgment call. I think trust gets built insight-by-insight starting with the former.
That's also partly why I'm deliberately keeping the Coach scoped to spending patterns and budgeting habits, not bigger financial decisions — for anything like investing, debt strategy, or major money moves, the app nudges people toward talking to an actual human advisor. The AI's job is to make the day-to-day stuff visible and actionable, not to be the final word on someone's finances.
Will report back once I have real numbers on the comprehension-vs-action gap — it's exactly the kind of thing I want to instrument for.
I appreciate you taking the time to explain that.
There’s an interesting distinction here that I’d like to discuss further outside the thread.
What’s the best email to reach you on?
Glad that was useful! Happy to keep talking — feel free to DM me on X, @coldboothq.
Thanks, I appreciate that.
I don’t use X for these conversations, but I’d be happy to continue over email. What’s the best email to reach you on?
Of course — smartspendrapps (this is a @gmail) works great, looking forward to it!
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.