REVERSIBLE Labs

The commerce graph for AI shopping. Product discovery, retailer offers & affiliate monetization across 700+ premium retailers. One API.

[REVERSIBLE Labs @reversible_labs]

@rdominguezibar Great Clueless thread. What still trips up most closet apps 30 years later is the jump from "outfit on screen" to "buy the missing piece in my size." Uploading the closet got solved. Live inventory across retailers mostly hasn't.

[REVERSIBLE Labs @reversible_labs]

@juansbiz Agreed, the demand signal before production is the interesting part. The other half is on the shopping side: once someone loves the look on their avatar, their size has to be in stock somewhere. That's where most styling apps lose the sale.

[REVERSIBLE Labs @reversible_labs]

@Jennywang @ForerunnerVC @eurie_kim @MenloVentures @LVMH @FortuneMagazine @FortuneMPW Congrats Jenny. $1B of closet inventory a month is the part that stands out: once you know what people own, the wishlist-to-checkout step gets a lot more precise. Curious whether designer partners will lean into that data.

[REVERSIBLE Labs @reversible_labs]

@MamundSajjad Shipping and returns gaps are real. The other one I'd score is variant data: agents often match the product but not the shade or size, and in beauty that's the whole purchase. Does Cartiq check variant-level availability?

[REVERSIBLE Labs @reversible_labs]

@RubanBhatia Makes sense, and it cuts both ways: if badges move the agent that much, clean data (exact variant, real stock, true price) should move it too. Do you know if the paper tested data quality, or only persuasion cues?

[REVERSIBLE Labs @reversible_labs]

@MisterSnowCrash Good test. I'd add the step before it: confirm the exact variant and live stock right before checkout. A lot of 'wrong item' cases in fashion are a size or colourway that changed between recommendation and purchase. Cheaper to prevent than reverse.

[REVERSIBLE Labs @reversible_labs]

@kl_amit Curious whether ShopGym covers variant-level failures. In fashion the agent often picks the right product but the wrong size or colour, or one that sold out mid-session. That's where most cross-retailer flows break. How did Upshot do there?

[REVERSIBLE Labs @reversible_labs]

@Bonigala Agreed, and the store side is shifting faster than people think. The pain in fashion is less the blocking than the plumbing: a separate affiliate deal and feed per retailer. We're collapsing that into one API at REVERSIBLE Labs.

[REVERSIBLE Labs @reversible_labs]

@JD_eth Interesting that returns and shipping terms are the biggest gap. Variants are the other one: size and colour often aren't machine-readable, so agents pick the wrong one. We normalize that at REVERSIBLE Labs. Does variant data factor into your score?

[REVERSIBLE Labs @reversible_labs]

@Jahazaibahmad I'd check first whether agents recommend the exact variant you have in stock. A 60-item catalog can't keep up, so agents browse live and land on whoever has the size. Clean variant and stock data is the cheapest way to stay in the pick.

[REVERSIBLE Labs @reversible_labs]

@gregisenberg Stage 3 is already visible in fashion. Retailers that earn on sales want agents in, as long as the agent brings clean data and an attributable sale. Agents that scrape get blocked; agents that can be credited for the sale get the door opened.

[REVERSIBLE Labs @reversible_labs]

@omooretweets Worth adding to the dummy-account test: ask for a specific size and colour, then check that the variant is actually in stock where the agent sends you. Showing up is step one. Being buyable at the moment the agent picks you is step two.

[REVERSIBLE Labs @reversible_labs]

@J_Akello Congrats on shipping it! The hard part in fashion discovery is usually the data: the same item listed differently by each store, and sizes selling out after the agent recommends them. We built REVERSIBLE Labs for exactly that. How is yours handling variants?

[REVERSIBLE Labs @reversible_labs]

@barsoapang With a text-first stylist, every extra shopping link is friction. Does Oro hand off to one buyable recommendation or several choices? Keeping wardrobe context alongside current retailer availability and delivery details could make that handoff much more useful.

[REVERSIBLE Labs @reversible_labs]

@tryfitsai The next useful step after 'would I wear this?' is 'where can I buy the exact piece?' We build fashion product matching and retailer-offer connectivity at REVERSIBLE Labs. Are you exploring that handoff, or keeping Tryfits focused on visualization?

[REVERSIBLE Labs @reversible_labs]

@neeraj343 The in-store Fit Check discovery is a useful roadmap signal: the same feature may be helping someone decide what not to buy. Did that change what you measure, from app engagement to whether a shopper felt confident enough to keep or reject a piece?

[REVERSIBLE Labs @reversible_labs]

@RobinWBH A useful first test: import five links from different affiliate networks, then compare the redirect chain and tracking parameters before/after. Preserve existing links unchanged; offer retailer discovery as an opt-in addition, not an automatic replacement.

[REVERSIBLE Labs @reversible_labs]

@medeana Could AI Shelf also score purchasability, not just brand mentions? For white tees: exact item and size, current stock, and delivered cost in the buyer’s region. That separates “the model recommends Uniqlo” from “the shopper can buy the recommendation.”

[REVERSIBLE Labs @reversible_labs]

@dipenvaja A useful spec is a quote the shopper can inspect: exact variant, retailer, delivered cost, return terms and when stock was checked. Revalidate before handoff; if price or availability changed, ask again rather than silently substituting.