Unique Research · Insight / Unique Friend Q&A · October 10, 2024 · Guest: CEO, Extreme Insight / Jirui Technology (极睿科技)
Jirui was later welcomed as Unique Partner (EN package 20241231a03ann00); this piece is the Unique Friend Q&A — inventory type remains insight.
Product / market: In content-coupled e-commerce, quality content drives GMV — but text, image, and video still burn people and time. As e-commerce dividends shrink and margins fall, merchants chase lower cost and higher efficiency. Jirui has completed four rounds led by GSR, Sequoia, and peers — nearly $200M raised; 1,000+ brand customers (Anta, Eifini, Schneider Electric, etc.); 40,000 stores covered; 2024 revenue expected >$80M.
Experienced sellers want cost cuts; inexperienced sellers struggle to guarantee content quality. AIGC automates polished marketing assets so all sellers get time-/labor-/mind-saving service and GMV from better content.
Four SaaS tools (each can cut e-commerce content cost ~80%; sold alone or bundled): photomagic (model-image synthesis); ECPro / Easy Merchandise (auto main images & detail pages + multi-platform publish); iCut (automated live-stream clips); iClip (image-to-video).
Focus: general e-commerce sellers. Tmall alone >1M sellers; domestic e-commerce content production ~¥100B/year — and a global need. Partners: Anta, Eifini, Schneider, Weilong, Estée Lauder; agencies Baozun, Baiqiu. Peak single-customer ARR >¥2M/year (Anta). 2024 expand cross-border/overseas.
Innovation: in-house AI for full-dimension image/video understanding; ECGPT + FashionCLIP for high-quality content; API one-click publish to ~20 major domestic/overseas platforms; finest-grain recognition + knowledge bases for structured non-standard goods.
GenAI impact: LLMs upgraded internal quality — text and video, not just copy/edit. Versus rule/template small models, large models yield more diverse, non-repeating brand assets. Pricing unchanged; GTM changed: (1) brands’ AI appetite spiked → many more inbound inquiries; (2) internal LLM-powered acquisition via massive short-video distribution and own live clips for leads.
Tech lead: (1) Stay on the front line — industry data is the moat as models commoditize. (2) Co-explore video and other models with foundation vendors (Tsinghua-rooted team, strong vendor ties). (3) Turn model edge into productized engineering — TOB customers buy tools, not raw models; ship full-chain e-commerce solutions against pain points.
Ecosystem: Model vendors ≈ cloud infra — not end-customer facing; apps sit in the middle. Joint custom solutions for bespoke AI needs; standardized products for standard needs; agency/human services downstream where AI+tools cannot yet finish the job.
Challenges / markets: China AI’s sharpest issue is value–price mismatch plus race-to-bottom pricing under crowded competition — customers equate AI with cheap automation. Hope: more diversified founder bets and a B2B culture closer to the U.S. that pays for value, not only raw cost.
2024 watch: Multimodal models (text→image→video) that remake expression in e-commerce and media. On-device LLMs that serve consumers at the edge — shopping may shift from search/content-driven to AI-guided / AI-decision e-commerce, changing platform and seller ops.
— END —