Why I’m Chasing an Idea That Doesn’t Scale
Someone had to save finance teams from Excel PTSD even in the age of AI
Updated
In a world where AI can code entire apps, you'd think getting business-specific reports would be a breeze. Unfortunately, it isn't.
The irony of the AI era is that while many finance tools now ship their own agents and copilots, most finance teams still deal with CSV exports, data cleaning and half-broken pivot tables regularly. As one June 2026 headline put it: finance teams can't quit Excel.
I've spent years in finance, accounting, built a strategic finance platform, and kept seeing the same ugly problem holding businesses back. Reporting workflows are a messy combination of humans dealing with real-world idiosyncratic business logic, and automations that assume idealistic data cleanliness. Not pretty.
"I love spending hours wrangling with data before I get to the actual analysis I signed up for," says no finance leader ever.
Consulting Meets Code
Finance leaders expect their bespoke reporting needs to be solved in this day and age. I'm talking about furniture companies wanting to see profitability by projects, F&Bs wanting to understand how inflation is hitting them ingredient-by-ingredient, advisory firms wanting a practice-branded report for all their clients but with different parameters per client. These messiest yet most insightful metrics typical software struggle with.
That's not a dig at typical software. It's just the reality of business – modern businesses juggle with delighting customers, capturing market opportunities, outmanoeuvring competitors, statutory admin, and optimising cost. Ensuring downstream data cleanliness and availability for potentially meaningful analysis is a luxury. Expecting businesses to operate around subsequent analytic needs is idealistic at best, and tone-deaf at worst.
What's changed isn't the problem – it's the economics. It no longer takes a team of engineers and a VC round to build something delightful. It is now possible for one person to provide concierge-level consulting services, build software and deliver post-sale support. The cost to build has collapsed, and with that, user expectations have skyrocketed.
Finance teams don't care how you automate their reports and dashboards, they just expect it to be done reliably. In this case, by a consultancy-software hybrid where a consultant/product builder (aka: me) goes deep into understanding the organisation's business logic and builds automated reports as if I'm part of the team.
"... but this doesn't scale"
Says everyone I've talked to about Cheetah. And they're absolutely right. Cheetah isn't designed to scale like a typical investable SaaS.
It is built to be deliberately small so I can move fast, ship fast, and learn fast. Scaling can wait. Right now, users are saying they want reports made to spec for their organisation and use case, and that's what Cheetah is focused on delivering.
Personally, I'm way more passionate about solving the most complex reporting needs than building scalable software nobody uses. Nothing satisfies me more than seeing users light up when the "impossible" finally works.
Update, September 2026: The core hasn't changed. Every report is still built to your spec around your business logic. What has changed is that the parts that repeat no longer need a person. Reports can now run on a schedule and land in your inbox as a Google Sheet link with a one-click Excel download (Scheduled Reports), and Cheetah no longer needs any access to your Google Drive (here's why).
"What's stopping someone else from doing the same?"
I'd like to think that having built hundreds of reports for businesses spanning across F&B, retail, real estate, finance and software over more than seven years of my career gives me some semblance of a moat.
Being able to speak the same language as CFOs and finance teams, and knowing how to robustly automate reporting micro-tasks definitely help too.
That said, I'm fully cognisant that this isn't rocket science. Anyone who understands the nuances and building blocks of financial reporting can do the same. There's probably someone out there who's reading this and considering to build an AI-first version of Cheetah.
That would be my goal at some point. For the foreseeable future though, getting users their perfect reports through human thoughtfulness precedes scaling (see above: Cheetah is not an investable SaaS).
The AI-first scenario is also less hypothetical than when I first wrote this: Xero now offers its own agent builder, and general-purpose agents can assemble a one-off consolidation. I've written about where that works and where it stops: recurring reports still favour something built to spec.
Software is dead, long live Software
My two cents as an investor-turned-product builder is that software isn't going away, users are just expecting way more from it. What the industry used to think of as scale – software that's built once, delivered ad infinitum at near zero marginal cost – is now seen by users as rigid, monolithic software that's tone-deaf to their unique needs.
The future of software isn't scale, it's specificity. Businesses want partners who understand their logic and automate around it. That's what I'm building with Cheetah. Deliberately small. Ridiculously nimble. Built to fit your business logic, and not force you into software that is optimised for VC's vision of scale.
Updated 20 September 2026: refreshed the opening on AI and finance teams' reliance on Excel, added an update on Scheduled Reports and Cheetah no longer needing access to your Google Drive, added a note on Xero's new agent builder, and changed "seven years" to "more than seven years".
A finance professional turned product builder, Jarvin has built hundreds of reports by hand and knows what financial and operational reporting demands: customisability, auditability, scalability, and security. Having automated that work reliably, he's now helping advisory firms and finance teams do the same.
