In Review · Shopify App Store
Deadstock AI — Shopify Inventory Recovery
A Shopify app that identifies dead inventory and surfaces one-click recovery actions powered by AI.
- Industry
- E-commerce / Shopify
- Engagement
- 4 months
- Shipped
- Apr 2026
- Role
- Nikshit Sehgal
In short
Deadstock AI is a full Shopify app — built by Brioodev — that scans a merchant's inventory, scores Cash-at-Risk on slow-moving SKUs, and surfaces AI-generated recovery actions (listing rewrites, bundle proposals, margin-safe discounts). Each action is a one-click approval. Currently submitted to the Shopify App Store and in review. Public launch follows approval. Built on the Shopify API, with the AI generation engine reused from SpecIQ's prompt-versioning architecture.
The Problem
What was breaking.
Past a certain inventory size — usually around 200 SKUs — every Shopify merchant has dead stock. Products that sold well two seasons ago, products that were over-ordered, products with sizes or variants that never moved. The cash sits on the warehouse floor instead of in the bank account.
Merchants know it's there. The reason most do nothing is that recovering it is genuinely tedious. You have to identify the slow movers, decide whether to relist, rewrite, bundle, or discount, then execute each action manually across the storefront and inventory system. For a merchant with 1,500 SKUs and three slow-mover categories, that's a week of work — every week — to do well.
Most merchants do the work badly or not at all. The cash stays trapped.
What Was Built
The delivered system.
Deadstock AI replaces the manual cycle with a structured pipeline.
Detection: the app scans the merchant's inventory daily, scoring each SKU on a Cash-at-Risk metric that combines stock value, time-since-last-sale, seasonality, and category benchmark. Slow movers surface to a triage view ordered by cash impact.
Generation: for each flagged SKU, the AI suggestion engine proposes three recovery actions — a listing rewrite (better description, better keywords, better images if available), a bundle proposal (paired with a high-velocity SKU), and a margin-safe discount level. The generation system is the same prompt-versioning + evaluation architecture we built for SpecIQ, adapted for retail.
Execution: each suggestion has a one-click approval. Approving a listing rewrite pushes the new copy to the product page. Approving a bundle creates the bundle in the storefront and schedules promotion. Approving a discount applies a price drop with a configurable end-date.
Tracking: every approved action is tracked — was the SKU's velocity higher in the 30 days after? How much trapped cash actually moved? Merchants see ROI on the work the system did, not just the work they approved.
The app is currently in Shopify App Store review.
Status
Where this stands today.
Submitted to Shopify App Store — currently in review. Public launch follows approval.
Tech & integrations
The tools this system runs on.
Named here for transparency — every choice was made for fit, not familiarity.
- Shopify Admin API
- Shopify App Bridge
- Next.js
- React
- TypeScript
- PostgreSQL
- OpenAI API
- Webhooks
- Vercel
What made it hard
Lessons we'll carry forward.
The hard part wasn't the AI — that pattern was already proven on SpecIQ. The hard part was the Shopify API surface area. Inventory data, product data, order history, fulfilment status, and storefront content live in different APIs with different rate limits and different freshness guarantees. Building a single coherent view of 'is this SKU dead?' across those surfaces was the bulk of the engineering. The lesson, which generalises beyond Shopify: when you're building on top of a platform, the platform's data model is the real product surface — not your UI. Treat it accordingly.
Have something similar?
Let's scope what you're building.
Discovery call first, scope document second, build third. We take 1 to 2 new projects per month.