Restocked

Know what pops before you buy.

The buying agent for independent shops. Paste a supplier's brand list, tell Restocked where your shop is and what your customers already love, and every brand comes back sorted into stock, consider or skip, with the evidence behind each call.

Live app · One-click demo · Sample shops · Pricing · MCP endpoint

---

Contents

• For judges: the 60-second path
• The problem
• What it does
• How it works
  • Architecture
  • One drop, call by call
  • How a brand gets its call
  • The agent's planner
• Why it needs Qloo
• Does it hold up?
• Use it from your AI assistant
• Project map
• Run it locally
• Limits, honestly

---

For judges

Open /app/demo. No sign-up. It loads a real drop for a Denver gift shop, recorded from Qloo's API.

1. Click any row in the table to see why the brand landed there.
2. Your call: tick, question or cross on any row to overrule the agent.
3. Order turns the final calls into an email to the supplier's rep, sent from the owner's own mail app.
4. Map shows where a brand is strongest across 12 US cities.
5. Qloo calls lists every request behind the drop.
6. Ask the agent on the right: "Why skip Baggu?", "What if I opened in Austin?", "Who buys Stanley?"

Criterion	Where to see it
Technological implementation	Seven Qloo capabilities used together: search, insights by taste, insights by city, explainability, demographics, audience compare and similar brands. Taste and place are asked as separate questions and combined by position (lib/grade.js). A paced, allow-listed Qloo client (lib/qloo.js). An MCP server with 5 tools. 66 tests.
Design	A full workspace: set up the shop → paste a line sheet → watch the agent run → drop table → your calls → order draft → share link or CSV. Light and dark themes; works on phones.
Potential impact	585,925 US retailers with fewer than 20 staff, selling $915B a year, commit cash to stock months ahead on instinct. The cost of inventory is their #2 problem out of 75 (NFIB).
Quality of the idea	Qloo on the supply side of retail: deciding what a shop buys, not what to recommend to a shopper. Checked against real stockist lists, three real shops and an LLM alone, with every result published (below).
"Would it work without Qloo?"	No. One supplier list sorted for four shops: a bestseller list gives all four the same answer; with Qloo, 21 of 23 brands get a different call depending on the crowd.

---

The problem

Independent shops place orders months ahead, usually on instinct and a sales rep's word. Sales history can't help with a brand they have never carried, and a bestseller list knows nothing about their customers. A wrong buy becomes dead stock and tied-up cash.

585,925	US retail businesses with fewer than 20 employees, selling $915B a year (US Census SUSB 2022)
3.2M	retail businesses across the EU with fewer than 20 employees (Eurostat 2024)
#2 of 75	cost of supplies and inventory, among small-business problems (NFIB 2024)
+19%	year-on-year growth in wholesale buying by independents (Faire, Q2 2025 → Q2 2026)

The cost of buying wrong: retailers lose $1.73T a year to inventory distortion, $572B of it overstock (IHL Group 2025, via Board), while the median small retailer holds just 19 days of cash in reserve (JPMorgan Chase Institute).

Why this becomes a big company

Tailwind
65% of UK retail is independent; independents are over half of retail locations in every US state	Faire, Jul 2026
605 new US independent bookstores opened in 2025; ABA membership up 151% in six years	ABA, May 2026
$1T of US retail revenue could come through AI agents by 2030	McKinsey via Digital Commerce 360
600M+ entities and 30T+ connections in Qloo's taste graph	Qloo

以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗