Working prototype · funding roadmap
Every squishy has a story.
Let’s give it a record.
Squishy Star is building the NeeDoh price guide that grades condition and shows every sale: careful identification, original and permissioned reference photography, completed-sale research, and values that show their work.
The prototype is live. Founding beta valuations are free, nothing is sold on the site, and every identification still requires human review.
Live evidence
What exists today
These figures come from the production research database. They are progress measures—not inflated coverage or accuracy claims.
Photo intake
Owners submit multiple views, packaging clues, visible text, and barcode details through a private upload.
Candidate matching
Server-side visual analysis extracts observable clues and ranks records already present in the Squishy Star database.
Human confirmation
A reviewer confirms the product and variant—or leaves it unidentified. The scanner cannot approve its own answer.
Evidence valuation
Completed sales, condition, packaging, and match quality produce a traceable range only when evidence is sufficient.
What crowdfunding unlocks
From a working foundation to a full NeeDoh price guide
Broader verified coverage
Research exact products, colorways, sizes, packaging, SKUs, UPCs, releases, and retirement status. Unknown facts stay visibly unknown.
A consented visual corpus
Expand original photography and opt-in, owner-reviewed private reference examples without turning personal uploads into public advertising.
Defensible value ranges
Collect completed-sale comparables, normalize shipping and lots, document condition, and keep active asking prices out of sold-price calculations.
Known and unknown benchmarks
Lock expected answers before each model run, measure Top-1 and Top-5 results, and publish failure rates—including confident wrong matches.
First public-release evidence gate
See the original photo library →100 products · 50 exact variants · 50 photo-backed variants · 25 UPC-backed variants · 10 products with sufficient completed sales · 100 labeled benchmark cases including at least 20 unsupported items.
AI and data transparency
Human judgment stays in the loop.
Squishy Star uses OpenAI’s server-side Responses API to extract visible clues and help rank candidate records. It does not train a private model on a person’s photos, automatically authenticate collectibles, or calculate value from appearance alone.
The reference database combines direct manufacturer facts, manually reviewed market evidence, Andrew Rider’s original household photography, and optional owner-submitted examples retained only with explicit consent. Private uploads stay protected. A contributor may decline reference-library use or request removal.
What the system will not claim
- A confident image match is not proof of authenticity.
- An active listing is not a completed sale.
- A product without sufficient evidence does not receive a fabricated value.
- An unsupported item remains unidentified.
- Campaign examples are not presented as independent customer traction.
Honest risks
What still has to be proven
Catalog coverage. The squishy market contains changing assortments and products with weak archival records.
Photo quality. Loose items, missing packaging, glare, and near-identical colorways can prevent exact identification.
Sale evidence. Some products trade too rarely for a defensible market range.
Human throughput. Trust requires review, source checking, and corrections; responsible growth cannot be instant.
Help build the collector record.
Try the prototype, keep the package, photograph every side, and let uncertainty be part of the evidence.
Submit a squishy for review