# HeyGen AI Startup Playbook — Riding the Wave to $100M ARR
**Date:** 2025-10-17
**Location:** Beijing
> *Joshua Xu*: **"Uncertainty is our advantage."**
## Milestone Achievement
HeyGen founder **Joshua Xu** announced on X (Twitter) that this month the company has reached **$100M ARR**. Remarkably, it took only **29 months** to grow from $1M ARR to $100M ARR.
---
## Core Philosophy
**Speed is Everything.**
HeyGen focuses on **leveraging AI advancements** in real time:
- Build products around constants (user workflows, core pain points).
- Design systems to benefit immediately from model improvements.
Joshua publicly revealed the *AI Startup Playbook* — the methodology behind how the company:
- **Thinks**
- **Collaborates**
- **Makes decisions**
- **Keeps release speed 5× faster than competitors**
**Original Playbook:** [https://x.com/joshua_xu_/article/1978837502787219578](https://x.com/joshua_xu_/article/1978837502787219578)
---
## HeyGen’s Mission
**Enable everyone to tell stories visually.**
### Types of Video:
- **Communication Videos**
Goal: clearly convey information. Edited with scripts.
- **Cinematic Videos**
Goal: inspire or entertain. Edited on timelines.
**Focus:** Democratize communication video creation for **all skill levels**.
---
## 1. Embracing Uncertainty
### 1.1 From “Solid Ground” to “Surfing the Wave”
- AI tech shifts disruptively every few months.
- Build strategies on constants, adapt products to unstable foundation layers.
| **Traditional Era** | **AI Era (HeyGen)** |
|---|---|
| Long-term tech stability | Continuous tech disruption |
| 12–18 month planning | 2-month wave cycles |
| Perfection before launch | Launch to learn |
---
### 1.2 Beyond Agile
Agile assumes technology stability. HeyGen optimizes for **rapid foundational change**, not small incremental delivery.
---
### 1.3 The Surfer’s Advantage
- **Identify constants** → build around them.
- **Leverage changes** → design products to auto-upgrade with model advances.
---
### 1.4 Speed with Quality
**Speed = faster user value + faster learning**
- Competitors release 1 feature/month → HeyGen tests 5.
- Learning compounds into massive product advantage.
---
## 2. Iteration Rhythm — Two-Month Wave Cycles
### 2.1 Roadmap Approach
- Plan every 2 months to match AI upgrade cycles.
- Daily releases.
- Biweekly deliverable commitments.
- 6–12 month strategic horizon.
---
### 2.2 Surfing Experiments
**Quick cycle example:**
1. Define hypothesis & criteria.
2. MVP in 1–2 days.
3. Small user test.
4. Analyze & decide.
**Good**: fast, data-driven, big bets.
**Fail**: slow, no learnings.
---
### 2.3 Fast Decisions — One-way vs Two-way Doors
- **One-way**: Rare, high caution.
- **Two-way**: Common, test quickly.
---
### 2.5 Managing Technical Debt
- Assume future replacement.
- Encourage tech debt pay-off if it boosts speed & reliability.
---
## 3. Operating Principles
### 3.1 Speed Mindset
- **Days, not weeks**
- Launch experiments when unsure.
- Action > perfection.
---
### 3.4 Deliver User Value
Innovation must **solve real problems**.
Measure **average video quality per user**, not just quantity.
---
## 4. Team Collaboration
### 4.1 Roles
- **PM**: Decision lead, strategy, MVP creation.
- **Engineer**: Fast builder, architecture for iteration.
- **Designer**: Simplify complexity, uphold “grandma usability”.
- **Data Scientist**: Validate impact, advanced metrics.
---
**Prototype Workflow**
1. PM/Designer + Engineer build quickly.
2. Validate idea.
3. Designer refines to product standard.
---
## 5. Product Team
- Builds core capabilities.
- Absolute best experience — **zero bugs goal**.
- 5× speed and iteration advantage.
---
## 6. Growth Team
- Experimentation engine.
- Focus on **impact**, not just code.
- Bold, fast hypotheses, willing to fail.
---
## 7. Communication Principles
- Asynchronous-first.
- Minimize meetings.
- Direct feedback: critique work, not people.
---
## 8. Pitfalls to Avoid
- Perfect architecture obsession.
- Research paralysis.
- Illusion of tech stability.
- Consensus trap.
- Quality as delay excuse.
### 8.2 When to Slow Down
Non-negotiable quality lines:
- Bugs blocking learning
- Security risks
- Major UX harm
---
## 9. Winning Principles
1. Move fast.
2. Best experience.
3. Quality first.
4. Discuss fully, execute firmly.
5. When unsure, experiment.
6. Ride AI wave.
7. Innovate via integration.
---
## 10. The Only Constant is Change
- AI advances rapidly.
- Seek learning, not false stability.
- Products should improve automatically as AI evolves.
---
> **Act now, don’t overthink, ride the momentum.**
---
