Hi Axelera team and community! 👋
I work in retail and F&B commercial operations, so I spend my life around shelves.
And there is one question every brand and every retailer wants answered in real
time but nobody can answer affordably:
> Is my product there, in the right place, at the right price — right now?
Out-of-stock alone is estimated to cost the industry tens of billions a year,
and most of it goes undetected because shelf observation is still done with
clipboards and checklists. Field audits are slow, subjective and infrequent.
Price and promo execution is decided at HQ and forgotten at the shelf.
**So here's what I want to build with Voyager Wingman: ShelfPulse.**
A compact camera device placed in front of a category runs a computer-vision
pipeline on an Axelera system that turns every shelf photo into a structured
audit in seconds, answering four questions:
1. **What's on the shelf?** — every product detected and identified
2. **Where is it?** — position checked against the target planogram
3. **What does it cost?** — price labels read directly off the shelf
4. **Is it available?** — empty facings and OOS flagged instantly
Every scan outputs structured JSON (per-facing SKU, position, price, stock
status, confidence) plus a live annotated preview. Accumulated over time, scans
reveal sell-through per facing, planogram drift, promo lift and OOS patterns by
store and category — turning the shelf into the industry's most honest market
research instrument.
**Why edge?** Privacy (the image never leaves the store), cost (affordable
frequent audits), and reliability in real aisle conditions — no store broadband
needed.
**How Voyager Wingman helps me build it:** prompt-first, on purpose. My first
prompt to Wingman: *detect every product on a shelf photo, crop each region, OCR
the price label, classify each facing as in-stock / OOS / misplaced against a
reference SKU set, and output structured JSON with a live annotated preview.*
Then iterate: model selection from the Zoo, OCR tuning for shelf-label fonts,
planogram comparator, annotated dashboard. Exactly what Wingman is made for —
from natural language to a running pipeline in record time.
**The build is deliberately scoped to one category, end-to-end:** photo in,
structured report out — simple, clear and buildable inside the four-week window,
with the planogram comparator and trend history as stretch goals.
**Why me?** I work daily with the users of this tool (trade marketing, category
management, store operations), I already run real-world data pipelines that turn
field data into retail insights, and I'm comfortable iterating on AI/CV stacks.
I know the problem from the inside and I'll deliver a working demo, not a
mock-up.
Looking forward to taking flight with a Wingman. Good luck to everyone!Â
