business model

How Beli makes money

Two revenue engines on the same asset: a copilot that turns shopping intent into confident purchases. Currently 69 SKUs across 11 partner merchants.

🔗

Affiliate commission

7.1%

Average take on every purchase we hand off. Consumer pays the same price — merchants pay for the conversion.

🧩

Merchant widget core engine

$49–199/mo

Independent retailers embed Beli on their own storefront — a sales copilot trained on their catalog. SaaS + per-conversation usage.

📊

Intent analytics

phase 3

Aggregated, anonymized demand signals ("searches for 'waterproof' up 40%") sold as category reports — never individual user data.

Live funnel — real events from this build

69

Catalog SKUs

11

Partner merchants

2

Copilot searches

0

Checkout handoffs

Unit economics (consumer side)

  • Avg. order value via copilot$180
  • Avg. affiliate rate7.1%
  • Revenue per checkout$12.72
  • LLM cost per session (local 7B → any OpenAI-compatible API)~$0.004
  • Contribution margin~96%

Roadmap

  1. NowCopilot MVP on a curated catalog — 5 categories, affiliate handoff live.
  2. Q2Widget self-serve: merchants upload a catalog CSV, embed with one script tag.
  3. Q4Cross-session memory + price-drop alerts; analytics dashboard for merchants.

Why we win

Every conversation teaches Beli which constraints actually predict a purchase — that preference graph compounds. Generic chatbots bolt shopping onto a model; we bolted a model onto commerce: grounded catalog queries, explainable picks, and a checkout that pays for itself. The moat isn't the LLM — it's the query-log → preference → conversion flywheel.