1
2 Comments

I'm a financial analyst who couldn't manage his own money. So I built the budgeting tool that was missing.

I know how that sounds. I work with financial data professionally — and for years, I couldn't keep my own household budget together for more than two months in a row.

It wasn't for lack of trying. I went through all the "major" apps. They all do roughly the same thing — connect to your bank via Plaid, auto-categorize everything, give you charts. Sounds great on paper. In practice, two things kept happening:

The auto-categorization wasn't done the way I'd like. I didn't get any "actionable insights" at the micro-level from the fancy charts. And because everything was automated, I'd stop checking in. The app was "handling it." Except it wasn't — it was just collecting data I wasn't looking at.

So I'd go back to spreadsheets. Download my bank's CSV, build formulas, categorize manually. This actually worked — I understood where every dollar went. But it took time I didn't have. One busy month and I'd fall behind. Two months and I'd give up. Then start the whole cycle over.

No one had built the middle ground. You either hand your bank credentials to an app and let it do everything (badly), or you open a blank spreadsheet and do everything yourself (unsustainably).

So I started building OpBoard.io

What it does:

You export a CSV from your bank (every US bank lets you do this — takes 30 seconds). You drag it into OpBoard. It uses AI-suggested keywords to auto-categorize your transactions into your expense categories.

But here's the key difference: you define the keywords. You control the categories. You see every transaction and can override anything. The AI suggests — you decide. Next month, when you import again, your keywords remember everything. The more you use it, the less work there is.

No bank login. No Plaid. No one touches your accounts.

Why I think this works:

Budgeting apps fail because they remove you from the process. Spreadsheets fail because they dump the entire process on you. OpBoard automates the tedious part (pattern matching transactions to categories) while keeping you in the loop on everything else.

I've been using it myself for months and it's the first time I've actually stayed consistent. If you've been through the app-spreadsheet-give-up cycle, you know that's not a low bar.

Where it's at:

Live at opboard.io. Free beta — no credit card. Solo founder, built nights and weekends with Next.js and Supabase.

It's probably not going to replace YNAB for people who love YNAB. But if you wanted spreadsheet control without spreadsheet maintenance — or you just don't want to give Plaid your bank password — You'll find it useful.

Would love feedback from this community. Beta key: OPBETA-PH-50

on February 24, 2026
  1. 1

    Really resonates with the CSV workflow approach. I've been working on something similar for small business bookkeeping and the exact same insight applies — the middle ground between full automation (that miscategorizes everything) and manual spreadsheets (that nobody maintains) is where the real opportunity is.

    Curious about your AI keyword categorization — how are you handling the edge cases? Like when the same merchant name shows up for both business and personal purchases, or when bank descriptions are just cryptic strings of numbers. That's where I've found most categorization tools break down. The happy path works great but real bank data is messy.

    Also smart to avoid Plaid dependency. The "export CSV" workflow feels clunky but it solves so many trust and security objections. People are way more comfortable with that than handing over bank credentials.

    1. 1

      Thank you for the comment.

      For those edge cases, I advise/allow users to identify and define what those are based on what they see on their bank lines. There are "ugly" descriptions, but as long as it has a recurring component within them, they should be able to set up their own keyword.

      In OpBoard, AI Suggestions are more like for low-hanging keywords. I found that it gets chaotic when I let the AI suggests outside the strict rules I set. Until AI gets as smart as human in terms of cognitive intuition, I intend to keep it this way.

Trending on Indie Hackers
Why Indie Founders Fail: The Uncomfortable Truths Beyond "Build in Public" User Avatar 127 comments I built a tool that turns CSV exports into shareable dashboards User Avatar 97 comments $0 to $10K MRR in 12 Months: 3 Things That Actually Moved the Needle for My Design Agency User Avatar 76 comments I got tired of "opaque" flight pricing →built anonymous group demand →1,000+ users User Avatar 45 comments A tweet about my AI dev tool hit 250K views. I didn't even have a product yet. User Avatar 43 comments Your AI Product Is Not A Real Business User Avatar 34 comments