Dead stock is the silent killer of small businesses. Not rent, not marketing — the shelves full of stuff nobody buys, with your cash frozen inside them.

I kept seeing posts from shop owners about the same cycle: overbuy inventory, watch it sit, run out of cash for the stuff that actually sells, repeat. A small shop doesn't have a data team to flag this. They feel the cash crunch but can't point at the cause. So I built StockSentry — a free tool where you upload your inventory CSV and it tells you, in plain numbers, where your money is stuck.

What it does

Dead stock detector. It scans your inventory and flags SKUs that aren't moving, with the exact dollar amount locked in each one. Not a vague "you have dead stock" warning — a ranked list with numbers you can act on.

Discount-to-clear simulator. Pick a dead SKU and it models what discount clears it in 30, 60, or 90 days. Sometimes the math says: take the 40% hit now, free up the cash, reinvest in winners. That's a decision a shop owner can make in two minutes instead of agonizing for two months.

Viral demand spike simulator. This one is fun — and genuinely useful. What happens if demand for a product goes 10x overnight (a TikTok video, a local influencer)? It simulates the stockout, the emergency orders, and the margin impact, so you can plan instead of panic.

True restock calculator. Supplier raises prices 25%? The tool shows how much that shrinks what you can rebuy with the same cash — a shortfall calculator that turns "prices went up" into an exact number of units and dollars you need.

Natural-language Q&A. Type "which products are critical?" or "how much cash is locked?" and get answers straight from your data, with the reasoning shown.

How it works

The backend is FastAPI. Analysis is a hybrid: deterministic heuristics do the heavy lifting on inventory math (reorder points, aging, turnover), and an LLM layer handles the natural-language Q&A and briefings.

One thing I'm proud of: the tool is honest about which is which. Every AI-generated output is labeled — you'll see whether a number came from the deterministic engine or an LLM narrative, and the LLM briefing falls back to a clearly-labeled heuristic mode if the provider is down. No fake precision. In inventory, a fake number is worse than no number.

The story

I built this for the InfiniSynapse hackathon, but the motivation came from real complaints. Dead stock isn't a "business strategy" problem — it's a visibility problem. The data is already in the shop owner's CSV export; nobody was reading it. I just made the CSV talk.

The demo CSV is preloaded so you can see everything in thirty seconds without having your own data handy — the six critical SKUs, the spike scenario, the price-hike scenario. When you're ready, upload your real export and it works the same way.

Try it

Live: https://stocksentry-bice.vercel.app
GitHub: https://github.com/rossycoder/stocksentry

Free, no signup. Upload the demo CSV or your own and tell me: what would you want it to flag that it doesn't? I'd genuinely love feedback from people who've lived the inventory crunch.