The default sales layer for e-commerce.
Every Shopify store deserves a 24/7 AI salesperson. We're building the one that only charges when it works.
The Problem
Shopify merchants spend $29–$499/month on chatbots that answer questions but don't close sales. They pay before seeing results. If the tool doesn't work, they still pay. The incentives are misaligned — the vendor profits whether the merchant does or not.
The Model
PalUp is an AI sales agent that embeds on Shopify stores. It recommends products, handles objections, and closes sales. The merchant pays nothing upfront — we take 5% of AI-attributed revenue only. If we don't generate sales, we don't earn.
This is a business model moat, not a feature moat. Anyone can build a chat widget in 2 weeks. No one can switch to revenue-share without gutting their existing subscription revenue.
Unit Economics
| Revenue per merchant | $125/mo | 5% of ~$2,500 AI-attributed GMV |
| Cost per merchant | $5/mo | LLM + embeddings + infra |
| Gross margin | 96% | $120 profit per merchant |
| Blended LLM cost | $0.0037/convo | 75% never touch Claude (Tier 0+1) |
| Breakeven | ~15 merchants | Ramen profitable at $1,600/mo fixed |
Why the cost is so low: 4-tier LLM routing + prompt caching. 75% of conversations are handled by cache or Gemini Flash-Lite ($0.001/convo). Only 20% touch Claude Sonnet ($0.016/convo). This routing strategy is the difference between 72% margin and 96% margin.
The Moat
The moat is the data flywheel. Every conversation teaches the AI what sells, for whom, and how — for each merchant specifically. After 3 months, switching to a competitor means starting from zero.
- Month 1: 300+ conversations. Basic customer profiles, product preference patterns.
- Month 3: AI knows objection patterns, cross-sell pairings, brand voice nuances. Competitor at $0 can't catch up.
- Month 6+: Customer profiles span repeat visits. Returning-visitor personalization that no new entrant can replicate.
The widget footer (“Powered by PalUp”) is a zero-CAC growth channel. Every store that installs PalUp advertises to every shopper that visits. Merchants who see the AI working on a competitor's store install it on theirs.
Market
| Shopify GMV (2026) | $800B+ |
| AI-assisted capture opportunity (~5%) | $40B |
| PalUp take rate (5% of AI-attributed) | $2B ARR at penetration |
| Expansion beyond Shopify | BigCommerce, Wix, WooCommerce |
Illustrative assumptions pending validation with beta data.
Competitive Landscape
| Model | Focus | Per-merchant learning | |
|---|---|---|---|
| PalUp | 5% revenue share | Sales | Yes — compounds daily |
| Rep AI | $29–99/mo | Sales + support | No |
| Tidio | $29–399/mo | Support | No |
| Gorgias | $60–750/mo | Support | No |
| Zipchat | $49–499/mo | Sales + support | No |
Competitors can't switch to revenue-share without destroying their existing subscription revenue. They have no incentive to build per-merchant learning because they get paid monthly regardless of results. Both moats are structural.
Where This Goes
Phase 1: Shopify. Default AI sales agent for the long tail of merchants ($10K–$200K/mo GMV). Product-market fit + unit economics validation.
Phase 2: Multi-platform. BigCommerce, Wix, WooCommerce. Same model, same flywheel, new surface area.
Phase 3: The data compounds across the entire network. PalUp knows more about what sells online than any human team. That intelligence becomes the platform.
A Note from the Founder
“We started PalUp because no merchant should pay a $299/month tax just to find out if AI can help them sell. If we don't make you money, you don't pay us. Every architectural decision — the 4-tier routing, the data flywheel, the revenue share — exists to make that promise real.”
— John, Taipei, 2026