TL;DR - I built NutriFinder, a neutral place to browse and compare endurance-sports nutrition - gels, drinks, electrolytes, bars, chews - on the numbers that actually matter when you’re moving fast: carbs per serving, sodium, caffeine, price per portion. Not on brand marketing. It has a race-day fuelling planner, a brand-facing portal, and an LLM-powered Playwright pipeline that refreshes the data weekly. It’s early, indie, and live. This is the story of what it is, why it exists, and where it’s going - including the bigger idea underneath: an open data layer for sports nutrition, Open Food Facts for the stuff you put in your race vest.
I spend a lot of my weekends chasing finish lines. Somewhere between my first half marathon and my first ugly bonk at kilometre 32, I learned that endurance sport is only half about the legs. The other half is fuel: how many carbs you take in per hour, how much sodium you lose in the heat, whether that gel with caffeine is going to help you or wreck your stomach. And the more I tried to get that half right, the more I ran into the same wall. The information was everywhere and nowhere. So I built NutriFinder to fix it, and this is the story of what it is and where it’s going.
What NutriFinder Actually Is
NutriFinder is a discovery and comparison tool for endurance-sports nutrition. Energy gels, sports drinks, electrolytes, bars, chews, recovery products: the stuff you stuff in your race vest. Instead of trusting a brand’s marketing page, you browse by category and compare products side by side on the numbers that actually matter when you’re moving fast, like carbs per serving, sodium, caffeine, and price per portion. At its core it answers one very practical question: given what I need on race day, which of these products is right for me, and how does it really stack up against the others?
Why I Built It
The honest reason is frustration. When I went looking for a straight answer about what to eat during a long effort, what I found was a swamp. Every brand tells you their gel is the best gel. Nutrition data is buried in tiny print on the back of a sachet, formatted differently by every company, and almost never presented in a way that lets you put two products next to each other and just compare. Forums are full of strong opinions and weak sourcing. The good science exists, but it’s scattered across studies, blog posts, and the collective memory of people faster than me.
What was missing was boring and essential: a neutral place where the real numbers sit side by side, brand marketing stripped away. That’s the same instinct behind most of what I build - a number out of context is close to worthless, so the honest move is to show the comparison straight and let the reader decide, rather than crown a winner for them. I’m an engineer, so at some point the frustration turned into a side project, and the side project turned into something I actually use before my own races. If I needed it, I figured other runners and cyclists did too.
What It Solves, And The Features That Do It
The problem NutriFinder solves is comparison on real data instead of vibes. Everything in the product bends toward that.
You can browse the catalogue by category and filter down to what you care about. When you’ve got a shortlist, you drop products into a side-by-side comparison to see the trade-offs at a glance. For the classic “is this one or that one better for me” moment, there are dedicated head-to-head “vs” pages that pit two products against each other directly. And you can save the ones you like to your favourites, backed up to your account rather than lost in a browser tab.
The feature I’m proudest of is the race-day fuelling planner, which lives at planner.nutrifinder.it and at /plan. Instead of just showing you products, it helps you build an actual plan: how much to take, how often, based on the duration and shape of your event, and it tries to explain the why behind the numbers rather than just spitting out a total. There’s also a growing library of guides and leaderboard-style pages, so you can land on a good answer even if you arrive from a search engine with a specific question.
There’s one more surface that isn’t for athletes at all: a brand portal, a separate dashboard where the companies whose products I list can see analytics and the impact of their presence on the platform. That matters because a good database needs the brands engaged, not just tolerated.
All of this runs on a fairly ordinary but solid stack: an Angular frontend, a Flask and PostgreSQL backend, hosted on AWS with nginx in front. Regular readers know I spend most of my time here arguing for the boring stack - as little moving machinery as you can get away with. NutriFinder is deliberately more than that, and the extra weight buys things the minimal version wouldn’t: a proper relational model for products and their nutrition facts, a separate brand-facing dashboard, room for the pipeline to grow. The boring stack is a default, not a religion, and this is a project where the defaults didn’t quite fit.
The less glamorous hero is that data pipeline. Keeping nutrition data current across dozens of brands by hand would be miserable and error-prone, so an LLM-powered scraper built on Playwright collects and refreshes product data from brand sites every week. I try to treat the model the way I argue teams should treat any LLM they bolt onto a system: as one component in a pipeline that still needs validation and a human sanity check, not a magic box you trust blindly. That pipeline is what makes the comparisons trustworthy, and it’s where a lot of the real engineering effort quietly goes.
Where This Is Heading
NutriFinder is the consumer face of a bigger idea. Behind it, I’m building toward an open data layer for sports nutrition: a genuinely open database that other apps, brands, and coaches could one day build on. The reference point isn’t a slick startup, it’s Open Food Facts. NutriFinder is how athletes touch that data today, but the data itself is meant to outgrow any single app.
Concretely, that means a few directions:
- Deeper category coverage beyond the endurance core.
- A smarter, more personal planner, so it adapts to your event and your gut rather than handing you a generic table.
- Better search and AI-assisted answers, so you can ask a real question in plain language and get a grounded response instead of ten blue links.
- A steadily wider database, until it’s the first place anyone checks a supplement’s real numbers.
Where Things Actually Stand
Let me be honest about the stage, because founder blog posts love to oversell. NutriFinder is live and in active daily development, but it’s early and it’s lean. This is an indie project run on a shoestring, not a funded machine with a marketing department. Features ship, get refined, and sometimes get reordered based on what real athletes tell me. I’m beginning to raise a small pre-seed round to give the project the runway to grow up properly, but nothing about the product’s usefulness is waiting on that. What’s next is more categories, a deeper planner, and steadily better data.
Come Kick the Tires
If you fuel for anything longer than a parkrun, I’d genuinely love for you to try it. Go to nutrifinder.it, compare a couple of the gels sitting in your kitchen drawer right now, and build yourself a plan for your next race at planner.nutrifinder.it. Then tell me what’s wrong with it. Early projects grow on honest feedback more than on praise, and the fastest way to make NutriFinder better is to hear where it let you down.
I’m building this in the open, one race weekend at a time, and I’d like you along for it. If you’re working on something similar - a data-heavy product, an honest-comparison problem, a pipeline you don’t trust yet - let’s talk.