- The Runway Ventures
- Posts
- 🤯 How Cleve Scaled From 0 to 20,000 Users in One Year
🤯 How Cleve Scaled From 0 to 20,000 Users in One Year
A real founder breakdown of product mistakes, unusable UX, and the rebuild that unlocked growth.
Admond Lee
December 14, 2025
Hey Founders,
👋🏻 Welcome to The Runway Ventures (🎙️Podcast Edition) — where I interview founders and deep dive into their stories to help you become the top 1% founder by learning from their startup journey and mistakes with actionable insights.
Today’s story is about Ashvin Praveen built and scaled Cleve to 20,000+ users within a year alongside mistakes and learnings. Let’s get to it! 🚀
Today at a Glance:
🫡 1 Founder → Ashvin Praveen (co-founder of Cleve)
⚠️ 2 Mistakes → Building a product that was “too clever” to use
🧠 3 Lessons Learned → Trust intuition first, then use data to validate
🔗 The Runway Insights → PostHog: A simple guide to validating product ideas
💰 Southeast Asia Funding Radar → Airwallex raises $330M (Series G) at $8B valuation to establish a new HQ in San Francisco
🫡 1 Founder: Ashvin Praveen (co-founder of Cleve)
🌟 The Highlights
🇲🇾 Founded by Ashvin Praveen, Lizzie Tan, and Muhammad Fathy Rashad, Cleve is an AI-powered content and personal branding tool that helps founders and creators turn ideas and voice notes into high-quality social media posts, fast and in their own voice.
The Problem — 😮💨 Founders, creators, and professionals know content is important, but creating high-quality posts consistently is slow, mentally draining, and hard to fit into a busy schedule.
Writer’s block, scattered ideas, and generic AI outputs make it even worse.
The Solution — 🧠 Cleve is an AI-powered personal content workspace that captures your ideas (text or voice), understands your context over time, and turns them into high-quality viral social posts in your own voice — so you can create consistently without spending hours writing.
You drop in your ideas — by typing or recording a quick voice note.
Cleve stores that context, understands your style, and then uses AI to turn those raw thoughts into polished social media posts.
Over time, it gets better at sounding like you, so creating content becomes fast, easy, and doesn’t feel forced.
Ashvin, Lizzie and Rashad (founders of Cleve)
🫡 Founder Story
Ashvin Praveen and his co-founders didn’t start Cleve from a big idea — they started it from a pain they lived with.
While running a personal branding agency as university students, they saw how powerful content was, but also how exhausting and time-consuming it could be.
After experimenting with dozens of ideas at Antler Malaysia, they built Cleve to help people create content faster without losing their voice. Early versions were clunky and confusing, forcing tough product resets and self-doubt.
🔥 Then LinkedIn Unwrapped went viral, pushing Cleve into the spotlight and validating their instincts. Today, Cleve is the result of constant iteration, hard lessons, and founders who kept rebuilding instead of quitting.
 | 🏗️ From day one, Ashvin built in public. He shared progress on LinkedIn, and their own content marketing attracted a wave of early adopters who felt the same pain. 🚀 Within the first year, Cleve went from literally zero to over 20,000 users on the back of this early traction. The turning point of Cleve’s early journey came in late 2024. The team was still far from their user growth targets, and had this wild idea of building a “Wrapped” recap for LinkedIn. So they went for it. In 2 weeks, the trio hacked together LinkedIn Unwrapped, pulling four all-nighters to ship the massive feature in time. It went viral. |
We launched it just before Christmas, thinking we’d be thrilled if it reached 3,000 users. The result blew their minds. In the first week, it hit 30,000. By the second and third weeks, it had passed 100,000 users.
🌏 Users from over 100+ countries started generating and sharing their personalised “Year Unwrapped” analytics, from university students to the CEO of PepsiCo Latin America posting their LinkedIn year-in-review. Even Alex Hormozi commented on their LinkedIn posts.
📍 LinkedIn Unwrapped put Cleve on the map.
Over 100,000 people tried Cleve’s tool within a few weeks — a level of exposure a tiny startup could only dream of.
🌋 The Lowlights
In the early days, the product was just absolutely unusable. Most people couldn’t figure out how to use it.
The early versions of Cleve were rough — really rough.
🫠 Ashvin admitted the product was “absolutely unusable” at the start, with users struggling to understand the UX and dropping off fast. The team tried to be too clever, reinventing workflows instead of sticking to familiar patterns, and paid the price for it. Every small UX mistake became a hard lesson in how unforgiving users can be.
At the same time, growth wasn’t moving as fast as they hoped. By late 2024, Cleve was behind targets, and the pressure started creeping in. Not to say that building mostly alone, watching peers take safer paths, and carrying the weight of uncertainty led to real self-doubt — feelings most founders don’t post about on LinkedIn.
🤳 The biggest reality check came when they realised most users were on mobile, but the product wasn’t built for it. Complaints piled up, features broke on phones, and it became clear that incremental fixes wouldn’t cut it.
Cleve didn’t just need tweaks — it needed a reset.
🚀 The Comeback
💪🏻 The turning point came when Ashvin and the team stopped patching problems and made a bold call — rebuild Cleve from scratch.
With most users on mobile, they went all-in on a mobile-first Cleve 2.0, rewriting the entire codebase despite how risky it felt. It was scary, but they knew the product would never work long-term without this reset.
Along the way, Ashvin shifted how he built products.
💟📊 Instead of obsessing over dashboards, the team started leading with intuition and direct user conversations, using data only to validate decisions. That mindset unlocked faster iteration and clearer priorities.
Combined with lessons from the viral LinkedIn Unwrapped campaign, Cleve emerged more focused, more confident, and built for how creators actually work — on the go.
📌 Here’s the TLDR of Cleve:
2021-2022 — ✌🏻 Ashvin and Lizzie ran whacked (a small personal branding agency) during university, using LinkedIn content to land clients.
This experience revealed the impact of content marketing and the lack of good tools, seeding the idea for Cleve.
Oct 2023 — 👌🏻 The founding team (Ashvin Praveen, Lizzie Tan, and Muhammad Fathy Rashad) officially formed during the Antler Malaysia accelerator.
They brainstormed ~30 startup ideas and decided to build an AI tool for personal branding/content creation.
Jan 2024 — 🇲🇾 Cleve.ai was born.
The team created a landing page and waitlist, validating interest from individual creators and even garnering a few enterprise LOIs.
Cleve was envisioned as an “AI Content Workspace” to help users write posts in their own voice, faster.
Mid 2024 — 🧑🏻💻 Cleve v1 launched (web only).
Early adopters (mostly solopreneurs, coaches, small-business owners) signed up via Ashvin’s content on LinkedIn.
The team’s organic marketing got the first few thousand users, proving demand.
🫡 However, user feedback flags major UX issues — the interface is confusing and not mobile-friendly. The team pivoted its focus to prioritise individual creators (B2C) over B2B after seeing stronger traction on the consumer side.
Late 2024 — 📉 Growth stalled slightly as Cleve approaches ~20k users, falling short of aggressive targets.
Major internal challenges around UX (many users struggling with the product) and platform (70% using it on phones) became obvious.
Dec 2024 — 🚀 Viral “LinkedIn Unwrapped” campaign.
In a last-minute push, the team built a LinkedIn year-in-review feature in 2 weeks (pulling 4 all-nighters).
Launched just before Christmas, LinkedIn Unwrapped exploded: >30,000 users in the first week, 100,000+ by early January.
The campaign spread to 100+ countries and dramatically boosted Cleve’s visibility, ending 2024 on a high note, riding the viral wave.
Jan 2025 — 📱 Cleve 2.0 relaunched.
On Cleve’s one-year anniversary, the team rolls out a complete product overhaul: native mobile app (iOS/Android) with a redesigned, intuitive UX and new features.
They rewrote the codebase from scratch to address prior failures.
This mobile-first approach pays off, as it catered to the 60–70% of users who prefer mobile. By this time, thanks to the Unwrapped boost, Cleve’s user count has crossed 100,000 and the startup is brimming with momentum.
Mid-2025: 🔥 Post-comeback, Cleve continues rapid iteration.
The team expanded (adding a full-stack engineer) and maintained a monthly update cadence with new AI features and improvements.
Having learned to trust their intuition (backed by data), Ashvin and his co-founders steer Cleve with a renewed product focus and a growing community of loyal users.
The rollercoaster first year, from struggling UX to viral success, sets the foundation for Cleve’s next chapters.
📈 Success is never linear, and Cleve is the best example here.
It took broken UX, slow growth, self-doubt, and a full product reset before things started to click.
💪🏻 But by staying close to their users, trusting their intuition, and having the courage to rebuild instead of quit, Ashvin and his team turned setbacks into momentum — and built a product that actually fits how creators work in the real world.
Want to learn more about Cleve?
⚠️ 2 Mistakes
Mistake 1: Building a product that was “too clever” to use
We tried to go against the grain with UX, but we realised recognisability matters way more than we thought.
🫤 In the early days, Ashvin and team kept testing different UI… and accidentally built something users couldn’t even operate.
He literally said the product was “absolutely unusable” and most people couldn’t figure out how to use it.
The painful part?
Their “innovation” was making people work harder, not easier. They only started seeing progress when they made tiny changes like button colours and followed more standard UX patterns — suddenly users “got it straight away”.
✌🏻 UX first, UI second.
Mistake 2: Trying to “find the answer” in data instead of building real product intuition
I used to spend hours in Mixpanel trying to figure out what to build next.
😂 Ashvin used to go into Mixpanel and spend hours trying to figure out what to build next — because everyone says “be data-driven”.
But he realised if you don’t have a strong hypothesis first, data becomes a rabbit hole → too many charts, too many interpretations, not enough progress.
✌🏻 Intuition first, backed by data.
🧠 3 Lessons Learned
Lesson 1: Don’t be clever, be clear
Ashvin’s biggest product lesson was basically → stop making users think.
He learned that repeatability and recognisability matter — sometimes copying existing patterns is better because users don’t need a learning curve.
🌮 Key Takeaways:
👀 Steal UX patterns from what users already know
Users don’t want to learn your product, they want to get their job done.
Once Cleve started copying familiar UX patterns (standard buttons, expected layouts, predictable flows), usage improved almost immediately.
Not because the product got “smarter,” but because it got easier.
⚡️ Measure “time-to-first-win” (not just signups)
For an AI content tool: the win is “I got a post draft I’m proud of.”
Track how fast a new user gets there.
💟 Design for the emotion (not the feature)
Creators don’t want “AI writing”.
They want: “this sounds like me” + “I can post today.”
Lesson 2: Trust intuition first, then use data to validate
Ashvin’s updated product philosophy is super founder-real:
Start with intuition → team debate → user conversations → validate with data
🤳 And the biggest “intuition + data” decision they made? Rebuilding Cleve into a mobile-first product.
They saw users using Cleve on mobile even though it was a web app, with complaints like instability and getting logged out.
Voice notes were a key feature for busy founders, but web-on-mobile limitations caused recordings to drop.
60–70% of users were mobile-first, including paid users — so they rewrote the codebase and started from scratch for 2.0.
The problem wasn’t data.
The problem was starting with data without a question.
🌮 Key Takeaways:
1️⃣ Step 1 — Start with an intuition
Write your intuition as a sentence.
For example, “I believe users actually want X, not Y”.
If you can’t write it clearly, it’s not ready.
2️⃣ Step 2 — For a hypothesis
Bad hypothesis → “Users want better UX”
Good hypothesis → “If we reduce steps from 5 → 2, activation will increase.”
3️⃣ Step 3 — Look at data to validate/invalidate your hypothesis
Ask:
What metric would prove me wrong?
What behaviour would surprise me?
Ignore everything else.
4️⃣ Step 4 — Make decision based on confidence
If hypothesis is validated, then do A.
If hypothesis is invalidated, then do B.
Make decision based on 60-70% conviction.
If you wait until data is “obvious”, you waited too long.
Lesson 3: Use “small bets” to learn fast, and “big bets” when conviction is high
💡 One of Ashvin’s most underrated lessons was learning that not all ideas deserve the same level of commitment. Early on, Cleve ran lots of small experiments — UX tweaks, onboarding changes, growth features like Roast Me. These were cheap, fast, and easy to kill if they didn’t work.
But when the team spotted a real opportunity (LinkedIn Unwrapped), they didn’t treat it like just another test.
⌛️ They paused product development, stopped marketing experiments, and went all hands on deck.
Ashvin realised that big ideas fail not because they’re wrong, but because founders hedge and execute them half-heartedly.
🌮 Key Takeaways:
🎰 Classify every idea before you build it
Decide upfront if it’s a small bet (fast, reversible) or a big bet (high conviction, asymmetric upside).
Don’t mix the two.
📊 Earn the right to go all in
Big bets should be backed by clear signals — timing, intuition, user pain, and upside.
If you can’t explain why now, it’s not ready.
⚡️ Stop everything else when making a big bet
Focus creates quality.
LinkedIn Unwrapped worked because Cleve didn’t multitask its execution.
📈 Use big bets to create narrative (not just growth)
In crowded markets (especially AI), bold, well-executed moves don’t just bring users — they shape how people remember your company.
In short, experiment cheaply, commit ruthlessly.
🔗 The Runway Insights
PostHog: A simple guide to validating product ideas (Link)
YC: The only question that predicts startup success (after reviewing 8,000 YC applications) (Link)
How to ask for help and actually get it (Link)
The founder’s playbook for performance culture (Link)
A few thoughts on conviction (Link)
Why startups die (Link)
💰 Southeast Asia Funding Radar
Airwallex raises $330M (Series G) at $8B valuation to establish a new HQ in San Francisco (Link)
ChemLex raises $45M to accelerate chemical discovery for the pharmaceutical industry (Link)
Validus bags $30M (Series D) led by Khazanah to scale in Indonesia and Thailand (Link)
MetaComp raises $22M (pre-Series A) to scale stablecoin payment network (Link)
🤝🏻 Before you go: Here are 2 ways I can help you
Founder Office Hours: Book a 1-1 call with me, share your problems and questions, and I'll help you cut through the noise, avoid costly mistakes, and get clear next steps that actually work. I help early-stage founders with:
Validating ideas & building MVPs
Tech & product development
GTM strategy & fundraising
Finding PMF & growth hacks
Growing & monetising newsletters
Attract customers & investors by building a solid founder brand on LinkedIn
Promote your business to 18,000+ founders: Acquire high-value leads and customers for your business by getting your brand in front of highly engaged startup founders and operators in Asia.
💃 Rate Today’s Edition
What'd you think of today's edition?Your feedback helps me create better content for you! |
That’s all for today
Thanks for reading. I hope you enjoyed today's issue. More than that, I hope you’ve learned some actionable tips to build and grow your business.
You can always write to me by simply replying to this newsletter and we can chat.
See you again next week.
- Admond
Disclaimer: The Runway Ventures content is for informational purposes only. Unless otherwise stated, any opinions expressed above belong solely to the author.
Reply