---
title: "Unique Interview | Extreme Insight × Chipuxingtu CEOs: From Burning Cash to Making Money in AI SaaS"
author: "Unique Research"
sourcePublication: "Unique Research (FFCAP English library)"
originalPublishedAt: "2024-11-08T22:47:47+08:00"
canonical: "https://ffcap.cn/en/research/20241108a0acv500"
source: "https://ffcap.cn/en/research/20241108a0acv500"
language: "en"
cnSourceUrl: "https://view.inews.qq.com/a/20241108A0ACV500"
---

# Unique Interview | Extreme Insight × Chipuxingtu CEOs: From Burning Cash to Making Money in AI SaaS

_Unique Research · Interview · November 8, 2024_

Guests: **Wu Bin**, founder & CEO, Extreme Insight / 极睿科技 (AI × e-commerce content); **Ping Yuan**, CEO, Chipuxingtu / 智璞星图 (AI ToB commercialization advisor; ex-Megvii VP ToB, DataOcean, OrionStar, UISEE). Full podcast in CN package.

## Building a scientific commercialization system

**Ping Yuan:** ~8 years commercializing AI — Megvii ToB across finance/auto/phone/RE/campus/education/retail; DataOcean IPO path; OrionStar robots; UISEE AV. Chipuxingtu now coaches deep-tech 0→1 GTM.

Deep-tech founders often ride intros to ¥10–30M revenue in years 1–3 without pain — then the real choice: solutions shop vs. product company vs. tech supplier. Step 1: _productize the tech_. Step 2: _solutionize the product_ — find horizontal industry demand and sell at scale. Most clear step 1; step 2 is where companies become services shops by saying yes to every custom order and miss the market. Slow is fine; rework is fatal.

**Solutionizing method:** do subtraction. Two filters: (1) Is the ask in your tech envelope without crushing R&D ROI? (2) Is it industry-wide rigid demand — not one big-account innovation pilot? Big orders ≠ industryization. Find general scenes inside your capability.

**Playbook:** (a) one lighthouse customer proving rigid need + willingness to pay; (b) 2–3 mid-market customers proving universality; (c) cost the tech investment; then scale to a fast, ¥100M+, stable line. Megvii 2017 face-unlock panel: don’t build OA/visitor systems in-house — partner ecosystem when expensive eng hours aren’t economic.

## AI-native commercial awareness

**Ping:** Foundation-model sales are harder than 2016–17 scene AI — they rewrite enterprise operating logic; CTO/CIO alone can’t greenlight; buyers are mostly large enterprises. Adapt: UX-first design; data-quality obsession; open ecosystem + standards.

**Wu Bin:** Extreme Insight — AI tools for e-commerce creatives (main images, detail posters, “guangguang” seed videos). Modes: self-serve tools; agency ops; AI-driven MCN that sells goods. GTM: ~20 direct AEs; channel partners (Jushuitan, Leyan) with rev-share; event selling; personal IP (~200–300k e-commerce followers). Challenge: prospects know ChatGPT but ask why pay vs. free open tools — differentiation is the job.

**Product edge:** vertical e-commerce data + fine-tuned models (better than general ZhiPu/Kimi/OpenAI corpora for shop scenes); heavy engineering — workflows beyond prompts, publish into Taobao/Tmall, pull click/conversion feedback.

**Pricing:** evolved from tools-only to tools + CPS — if we add ¥10M GMV, ¥1M commission is fair. Software fee + sales commission mix.

## GTM & pitfalls

Outcome/CPS patterns travel when value is measurable; Ping stresses quantify cost→value before customer talks. Effective AI-native GTM: lighthouse → mid-market proof → partner amplify; avoid over-promising (both guests). Shared pains: custom sprawl, unclear unit economics, hiring sales before product-market fit. Wu: e-commerce ecosystem partners (Leyan et al.) more complement than agency competitors. Roadmap: deepen vertical models + closed-loop GMV attribution.

— END —

---

Editorial English edition published on UniqueCapital: https://ffcap.cn/en/research/20241108a0acv500
On-site reading page: https://ffcap.cn/en/research/20241108a0acv500