AI startups

Trickle Founder: Three PMF Milestones in AI’s Global Expansion | Linkloud Salon #34 (Part 1)

Honghao Wang

15 Oct 2025 — 4 min read

Linkloud Introduction

A breakout app attracting 1 million users within 48 hours might seem like the ultimate Product-Market Fit (PMF) signal — but for Trickle, it became a dangerous prosperity trap.

Before reaching $1M ARR in just one month with the recent Magic Canvas release, founder Xu Ming navigated a winding path filled with missteps and “wrong answers.”

At the latest Linkloud offline salon in Shenzhen, Xu Ming shared his complete journey of trial-and-error. He spoke about:

Why he abandoned a nearly year-long failed project that was "polishing something fundamentally broken."
How he found truly high-value customers through complaints, not viral traffic.

---

Key Takeaways

PMF is a long-term process — don’t rush it, except for rare “star products.”
Never fool yourself without PMF — fixing surface flaws won’t help if the foundation is wrong.
Context beats prompts in AI workflows.
Once certain things click, multiple factors will start working in your favor.
The hardest moment in many markets is when everyone else has already found PMF.

---

The Magic Canvas Breakthrough

Trickle’s release of version 2.0 — Magic Canvas — achieved 97% organic traffic and $1M ARR in one month. This came after multiple “exams” on Product Hunt, starting from the first launch on August 3, 2022, and evolving through three distinct phases.

---

Phase One: First Attempt — Collaboration Tool with a Flawed Foundation

During the pandemic, online collaboration tools thrived. Trickle’s team built an “internal social feed” for asynchronous team communication.

Problems:

Real-time tools already dominated the market.
Internal sharing wasn’t a strong need for most employees.
Engagement and log-ins were extremely low.

Critical mistake:

Rather than analyzing root causes or doing user research, the team added more productivity and chat features — trying to be both Notion and Slack.

Misleading PMF signals appeared when small teams without previous productivity tool experience adopted Trickle, but these users came from low-tech environments (often Excel users).

After a year of effort, the team turned to AI — but ultimately AI couldn’t save a broken product.

---

AI Insights That Endured

From this failed attempt, Trickle learned:

AI has huge potential in workflow scenarios.
Context matters more than prompt quality — AI responses are more accurate when tied to specific documents/tasks.

This led them to rebuild from scratch with a context-centric philosophy.

---

Phase Two: Small AI Tools Making a Comeback

Identifying Real Pain Points

Discovery: Most team members frequently sent informational screenshots to WeChat’s File Transfer Assistant.
Verification: Twitter and Reddit users worldwide complained about messy screenshot management.
Constraint: At the time, GPT-4 was text-only — making screenshot text extraction (via OCR + ML + GPT summarization) a unique workflow.

Within 2–3 days, Trickle had a functional prototype.

Streamlining the Product

Early feedback revealed image handling was the real need. Other features (notes, bookmarks) were cut.

Result:

A simple drag-and-drop image upload with AI processing.

---

Effortless Product Hunt Success

Posted candidly on Hacker News — pinned for 3 days.
Launched on Product Hunt the next day without marketing.
Ranked #1 organically after competitors were disqualified.

This attracted:

Organic traffic from newsletters.
Attention from firms like A16Z.
~400–500 paid subscribers on day one via paywall + 7-day card-linked trial.

Lesson:

Small, obvious unmet needs can turn into early PMF — especially when charging boldly from the start.

---

Phase Three: Finding the True ICP

Coding for Everyone

Early experiments allowed users to build apps in natural language — but barriers were too high, and models couldn’t code well yet.

Later, a mini-game called Color Find, built by a user, drew 1M players in 48 hours. Believing this signalled ICP, the team focused on fun web apps — but traffic from virals didn’t convert to loyal or paying users.

---

The Real ICP — Internal Tools & Personal Apps

Unexpectedly, enterprise IT staff and business department employees adopted Trickle for dashboards and internal boards — always using the built-in database.

Database usage proved a strong retention indicator. These users rarely coded but constantly used the apps — solidifying PMF.

---

PMF Lessons

Observe users deeply
Don’t build exactly what they ask for — understand the root cause.
Example: Users requested “file locks” for AI coding. The real cause was lost context, solved by Magic Canvas persistent pinning.
PMF accumulation
It’s often gradual — built through insight, iteration, and tracking behavior.

---

Take Action with AI-Powered Tools

For creators scaling their PMF-driven products, platforms like AiToEarn官网 can:

Generate AI content.
Publish across Douyin, Kwai, WeChat, Bilibili, Xiaohongshu, Facebook, Instagram, LinkedIn, Threads, YouTube, Pinterest, X (Twitter).
Provide analytics and AI model rankings (AI模型排名).

This end-to-end ecosystem supports faster iteration from idea to value realization.

---

Final Takeaways

Don’t push in the wrong direction — slow down when anxious.
PMF isn’t the end goal — in saturated markets, differentiation is key.
Start now: believe, act, and adjust dynamically.

We focus on global innovation trends — add our WeChat assistant if you share this interest!

---

Past Highlights:

---

For multi-platform storytelling without extra workload, AiToEarn官网 offers open-source AI content generation, publishing, and monetization — ensuring your story reaches both investors and customers effectively.