Evie | WeLike 🦾🍀

Founder of @WeLikeHQ, AI-native Community & Growth Partner ｜ 100+ projects GTM experience ｜ Building @JELabs2024 ｜ Ex @okx ｜ CN @0xEvieYang

[Evie | WeLike 🦾🍀 @eviegrowth]

@Davidjpark96 Congrats!

[Evie | WeLike 🦾🍀 @eviegrowth]

Don’t Let Overhyped ARR Change Your Growth Plan

Internet companies used to talk about users, DAU, and retention. AI companies now talk about ARR.

Every week, there seems to be another ARR milestone: $5M, $20M, $50M, $100M…

The numbers keep getting bigger. Sometimes it feels like an AI company needs an impressive ARR figure before it can do any PR.

ARR is not useless. For a subscription business, it is an important growth metric. The problem is that companies may calculate it very differently.

Some ARR numbers come from:

1/ Taking one strong month and multiplying it by 12
2/ Annualizing a short-term usage spike
3/ Counting short pilots as annual contracts
4/ Including signed deals before the cash arrives
5/ Treating one-time services as recurring revenue

These numbers may actually be run-rate revenue, bookings, contracted ARR, or services revenue. But in public, they are often all called ARR.

The final screenshot shows one big number—with no churn, no time period, and no explanation of whether the revenue will last.

For AI companies, the same $10M ARR can represent very different businesses.

One company may have customers using the product in production, renewing regularly, and generating healthy margins. Another may depend on short pilots, manual services, and high inference costs.

The number is the same. The business quality is not.

That’s why I would also look at:

1/ Pilot-to-production conversion
3/ Renewal rate, GRR, and NRR
4/ Cash collected
5/ Gross margin after inference costs
6/ CAC payback
7/ Customer concentration

While working with AI teams on growth, I’ve seen how a competitor’s steep ARR curve can create anxiety. Teams start changing roadmaps, increasing budgets, and spending more on acquisition to chase a growth curve that may never have existed.

Now, I see public ARR as a market signal—not an operating benchmark.

What matters more is whether users stay, customers renew, revenue turns into cash, and the unit economics work.

Growth is not a screenshot contest. Every team still has to build its own product and business.

Keep building.

[Evie | WeLike 🦾🍀 @eviegrowth]

Claude’s (@claudeai) "Artifacts" feature is seriously so GOOD.

Used to be, writing docs with AI meant: chat a round → copy → paste into my draft → hop back to revise. My thinking kept bouncing between the chat window and the document.

Artifacts are different. The doc opens right inside Claude. I can ask and edit at the same time: add a paragraph, rewrite a section, fill in the materials, and see the changes instantly. The model and I collaborate on the same draft, instead of passing files through a chat box.

For me, the biggest change isn’t better prose. It’s that the writing rhythm doesn’t break. Ask, write, revise: finally one loop, in one interface.

[Evie | WeLike 🦾🍀 @eviegrowth]

In GTM for vertical AI products like music and visual creation, opposition to AI can sometimes sound louder than enthusiasm. This is even clearer among art creators.

On one hand, AI inevitably touches the identity question of "who I am." For some people, protecting that sense of self matters more than efficiency gains or a wider capability frontier. On the other hand, critical voices tend to be more unified and more audible. Positive voices usually take far more effort to surface and sustain.

[Evie | WeLike 🦾🍀 @eviegrowth]

Content creator whose new content is apps" feels closest to what we're seeing in AI GTM: distribution-native founders who ship the product as the next post. One pattern: they already own attention in a niche, so PLG starts as a follow, not a cold landing page.

[andrew chen @andrewchen]

what are the new types of founders that might get unlocked in the post-AI era?

- solo founders
- non-technical founders
- domain expert with no tech experience (ex: a dentist builds dental SaaS empire)
- content creator whose new content is "apps"

... who else is on this list?

[Evie | WeLike 🦾🍀 @eviegrowth]

Some thoughts on the Kimi English-market posts that got aired by a “KOL”:

Setting aside the brief itself (talking points, paid labels, and so on), I think the real issue is the relationship with the creators.

When a brand and KOL collab is still a short trial and trust is thin, private materials are much more likely to get posted once talks fall through.

Assuming Kimi did have outreach with this creator: if the reach from going public beats the upside of keeping the relationship, it’s no surprise they went public. The looser the relationship, the less control the brand has.

Picking a KOL isn’t just metrics, content, or reach. Delivery history and character matter too. For creators who are often in controversy and hard to align with on boundaries, even strong traffic still needs caution.

That’s also why brands with little local experience or weak creator networks often go through an agency; and why more mature brands treat KOLs as more than a paid post. They keep building the relationship online and offline.

From the KOL side: if brand deals are part of your work, you’re also managing your commercial reputation. Saying no or disagreeing is fine. But airing materials from a deal that never closed, or that’s still private, can seriously hurt how much brands trust you.

Short-term, this creator got reach. Long-term: how do brands trust them enough to work together again?

One last note: among the big AI companies, Kimi’s marketing is still my favorite. Tasteful, measured. The campus recruiting session at the old temple yesterday showed real brand taste too. This English-market episode doesn’t change that. The flaws don’t outweigh the strengths.

[Evie | WeLike 🦾🍀 @eviegrowth]

The core of AI growth isn’t producing more content. It’s learning about your users faster.

This insight comes from Hector Serrano( @hectorserrranoo) Head of Content at Wispr Flow and the company’s second marketing hire.

He recently shared how Wispr Flow built its paid ads creator program from scratch—scaling from five creators to roughly 50 active creators in four months. Today, the team ships 500–700 pieces of UGC ad creative every month.

But the volume isn’t the most interesting part.

What caught my attention was how they integrated creators, creative testing, and paid acquisition into a single GTM learning system.

Here are four takeaways for anyone building AI GTM:

1/ Creators shouldn’t just be content suppliers. They should be growth partners.

Traditional creator marketing pays per deliverable.

The brand takes on the media-buying risk, while creators have little incentive to keep improving performance after the content is delivered.

An emerging alternative is to compensate creators based on ad spend: the more budget their creative earns, the more they make.

For example, if an ad receives $20,000 in media spend and the creator earns a 7% share, their payout is $1,400.

But there’s an important problem:

Ad spend ≠ revenue—and it definitely doesn’t equal LTV.

An ad receiving more budget doesn’t necessarily mean it is acquiring higher-value users.

The key to performance-based incentives isn’t simply tying creator compensation to media spend. It’s aligning creator incentives as closely as possible with real business outcomes.

------

2/ Recruiting creators is really a way to test GTM hypotheses.

Many teams start by finding creators, then decide what they should create.

A more effective sequence is:

ICP → Use Case → Creative Gap → Creator Selection

For an AI coding product, developers, founders, and general knowledge workers may have completely different needs for the same feature.

A niche creator who genuinely understands developer workflows may be far more valuable for validating a specific use case than a general tech influencer with millions of followers.

Creators don’t just help brands reach users. They help uncover demand that hasn’t yet been validated.

This is especially important for AI companies expanding globally.

Localization isn’t simply translating content. It means revalidating local users’ pain points, use cases, and willingness to pay.

------

3/ Creative testing isn’t about finding a winner. It’s about understanding why it wins.

At many companies, creators and media buyers operate as two separate systems.

Creators produce. Media buyers distribute.

When an ad underperforms, the team swaps out the hook or rewrites the script.

Without structured feedback, creative iteration quickly becomes guesswork.

A more mature approach is to build a shared creative taxonomy:

Persona × Concept × Angle × Awareness

Then analyze it alongside hook rate, hold rate, CTR, and downstream metrics such as activation, CAC, and retention.

If an ad performs well, was it because the hook captured attention?

Or because a specific persona has stronger demand for that particular use case?

Only by understanding the mechanism can a one-off success become a reusable growth insight.

Ad data shouldn’t only optimize media buying. It should also inform product positioning and user research.

------

4/ The biggest growth trap is endlessly replicating ads that already work.

When creator compensation is tied to ad spend, creators naturally gravitate toward imitating winning ads.

But if 50 creators are all copying the same concept, the result may simply be 50 different faces and 50 variations of the same hook.

Content output appears to multiply, but the company isn’t discovering any new growth territory.

This is the classic exploration vs. exploitation problem in performance marketing.

Proven creative is responsible for scale.

New concepts are responsible for learning.

They should have different budgets, evaluation criteria, and incentive structures.

Otherwise, the harder a team optimizes for short-term ROAS, the more likely it is to lose its ability to discover new markets, new users, and new sources of demand.

------

One final thought:

I’ve always believed that an AI company’s GTM team shouldn’t merely be a traffic-acquisition function.

It should be the market research team closest to real user demand.

This matters even more when AI products are evolving quickly and user behavior hasn’t stabilized. At this stage, every creative test is essentially a low-cost market experiment.

Producing 700 ads per month is certainly a sign of strong execution.

But the more important question is whether those 700 experiments helped the team understand:

Who will buy?

Why will they buy?

Which use cases create the most value?

Where is the next growth opportunity?

If the team hasn’t gained any new insight, even the most efficient creator program is still just a content factory.

The best AI GTM systems don’t simply help teams produce content faster.

They help the entire organization learn about its users faster.

[Hector @hectorserrranoo]

i built @WisprFlow’s paid ads program

here’s how i’d rebuild it in 30 days

[Evie | WeLike 🦾🍀 @eviegrowth]

The Secret to 5x User Growth ROI: Ambassador Programs

Kimi @Kimi_Moonshot recently launched its Global Ambassador Program. Beyond Kimi, top AI companies—including Anthropic, Codex, and others—have also been rolling out their own ambassador programs.

The reason behind this trend is actually quite simple. Anyone in marketing knows that relying solely on paid ads for user acquisition comes with two unavoidable pain points: costs keep climbing, and the impact is increasingly short-lived (metrics spike during the campaign, but drop the moment ad spend stops).

So the core challenge to solve is: How do we make product exposure and user growth sustainable while keeping costs under control?

For our team, the answer is: Building an Ambassador Program.

Since the first half of 2025, our team has been collaborating with partners to design and execute ambassador programs. In terms of results, through these programs we have achieved:
✅ Nearly $2 billion in cumulative trading volume for a financial/trading product.
✅ Millions of monthly impressions for an AI product, reducing overall CPM by ~60% compared to their previous ad spend.

——

3 Main Ambassador Collaboration Models
If you are considering launching an ambassador program, check which of these three common models fits your product best:

1️⃣ Retainer Fee + Content Tasks: Ambassadors receive a fixed weekly or monthly payment in exchange for a set volume of content output. Best suited for early-stage products that need rapid market saturation before conversion paths are fully validated.

2️⃣ Revenue Sharing / Commission: Ambassadors earn a percentage of GMV or revenue generated when referred users execute trades or buy subscriptions. Best suited for trading platforms or products with clear monetization paths.

3️⃣ Resource Exchange / Perks: No cash is exchanged; instead, ambassadors receive fee discounts, priority access, or exclusive resources for their teams. Best suited for strong brands where ambassadors genuinely want deep access to the product itself.

Note: In practice, real-world programs often combine these three models, frequently adding bonus prize pools for extra motivation.

——

Why Ambassador Programs Outperform Paid Ads in ROI?

Marketing appears to be about creating exposure, but at its core, it’s about building trust.

Paid ads buy attention, and attention can easily be stolen away. Ambassador programs build trust, and trust is far harder to steal. The ambassadors themselves serve as the bridge of trust between your product and your audience.

——

While ambassador programs sound simple in concept, they are extremely operations-heavy in practice.

From choosing the collaboration model to recruiting, screening, incentive design, and long-term retention—every single stage is packed with nuances. These details are often what determine whether a program becomes a scalable engine or a total flop.

In my next post, I’ll break down the specific pitfalls our team encountered and the playbook we developed while running these programs.

[Evie | WeLike 🦾🍀 @eviegrowth]

🔗 WeLike Social Listening: https://t.co/DvLA9w9QQs

[Evie | WeLike 🦾🍀 @eviegrowth]

Following KOL Pricer, our tweet pricing tool that hit 140K+ impressions, our team just launched our second super practical AI marketing tool: Social Listening!

Enter the X account of any product you're tracking, and within 30 seconds you'll get a full breakdown of what real X users are saying, including:

→ Sentiment analysis
→ Buzz trends with milestone context
→ Hot topics with suggested marketing moves
→ A market strategy report
→ AI Q&A

Want to monitor a product's social chatter in real time? Turn on "Smart Alert" and everything you're following gets pushed straight to your Telegram.

There are three reasons the JE Labs (@JELabs2024) team built this:

1/ Every time a client launches a big campaign, we watch social and community channels closely to gather user feedback and catch any PR issues early, so we can respond in time.

2/ When we design talking points for a project, we dig through user discussions to see how people naturally understand the product, what needs correcting, and what's worth amplifying. It gives us a solid foundation for the next iteration.

3/ We also track how discussion around a product trends over time, pinpointing the key moments that tell us what topics and features the market actually cares about.

For a long time though, we did all of this by hand. It was time-consuming, and the information we gathered wasn't always complete or timely. So we took the thesis from our own hands-on experience, broke down the information sources we follow and the angles we read them through, and built Social Listening to do it for us.

For marketers, it's a no-brainer tool for working faster. And for anyone trying to understand a product, whether you're a user or a researcher, it helps you get a read on real user feedback in just a few minutes.

WeLike Social Listening (@WeLikeHQ) is free to use right now. Go try it out and let us know what you think!

[Evie | WeLike 🦾🍀 @eviegrowth]

Calling all AI Agent builders, founders, and investors across the stack to our Friday Happy Hour.

AI Agent Builders Gathering brings together the full stack of the Agent ecosystem, from model layer to inference to frameworks to applications, for one night of real conversations.

No panels. No fluff. Just builders.

📅 May 8, 2026
🕕 6:00 – 9:00 PM
📍 San Francisco (exact location shared upon RSVP)
🔗 https://t.co/LEM0mvq1mR

[JE Labs @JELabs2024]

⭐️We’re co-hosting an AI Agent Builders event on May 8th in San Francisco!

We welcome AI Agent builders, founders, and investors across the stack to explore what’s next.

Don't miss:
- Real conversations
- High-signal networking
- Fresh perspectives on the agent stack

Co-hosts
@CreaoAI @GoKiteAI @YottaLabs @kuseHQ
Event partners
@JELabs2024 @CollovLabs @gptdaoglobal

Spots are limited
See you there

👉https://t.co/qS9BJqJUIP

[Evie | WeLike 🦾🍀 @eviegrowth]

o excited to find NUS alumni family in the Bay Area! 🦁🇸🇬

Even back when I was a student, NUS was a powerhouse for driving innovation and entrepreneurial programs.

Fast forward to today, stepping into the shoes of an alumna, I'm experiencing the university's incredible support for the AI and frontier tech ecosystem from a whole new perspective.

As more and more AI startups expand overseas and anchor their APAC operations in Singapore, I am beyond excited for the renewed energy and vitality flowing into the city 🫶🫶🫶

[Evie | WeLike 🦾🍀 @eviegrowth]

I’ve truly felt the gap between Vibe Coding and real product development:

With Vibe Coding, you can whip up a 70-point demo in just 3 hours.

But to bridge that extra 20-point gap and polish that demo into a finished product, it takes well over 30 hours.

That final 20% requires deeper industry know-how, sharper product sense, and much broader data sources...

[Evie | WeLike 🦾🍀 @eviegrowth]

Most projects are burning their hackathon budgets for minimal return.

After attending several recent events, my biggest takeaway is clear: Hackathons are an elite GTM channel for dev-tools, but only if you stop treating them as "offline-only" moments.

While they are unmatched for acquiring seed users and gathering raw feedback, the ceiling for an offline event is usually a few hundred people.

The real leverage? Digital amplification.

By bridging the gap between the venue and the feed, projects can 100x their ROI—turning 200 local impressions into 20,000+ global ones.

A few approaches I've seen work well:

1/ Pre-event Momentum: Don't just announce; build anticipation. @photonhq set a great example by shooting a high-production promo for their Residency, resulting in 5x their average engagement.

2/ Capture the Energy: Use on-site interviews, rapid-fire Q&As, and BTS clips to make those not in the room feel the FOMO.

3/ Incentivize Distribution: Make "online visibility" a judging criterion. When participants are rewarded for sharing their builds, every hacker becomes a distribution node.

4/ The Long Tail: Winning demos shouldn't die on stage. Feature them as case studies and amplify them across official channels to prove long-term utility.

Offline is for depth; online is for scale. Don't leave 90% of potential ROI on the table.

[Evie | WeLike 🦾🍀 @eviegrowth]

Congratulations to the Creao AI team on completing 3 rounds of funding within a single year, raising a total of $30 million! 🎉

Recently, @CreaoAI's product updates and iterations have been incredibly fast—their 20-person team shipped 8 new features in just one day.

Furthermore, the paid conversion rate among overseas users is multiplying, and their ARPU is 6 to 7 times higher than that of regular 2C AI tools.

I believe this is all just the beginning. Stay tuned…👀

[Kai Cheng @Kai900607]

We just raised $10M led by Prosperity7 (Aramco Ventures).
$30M total in under a year.

We got here by being our own crash test dummies.

On the engineering side, the team rebuilt our entire codebase around AI. 99% of our production code is now written by agents. We ship daily. Features go live the same day they’re conceived. Bad ones get killed the same afternoon.

On the GTM side, the same thing happened. Google Ads audits, GA4 breakdowns, SEO gap analysis, content pipelines. All running on agents. A 20-person team doing what would normally take multiples of that headcount.

We didn’t bolt AI onto how we work. We redesigned how we work around AI. Engineering, product, marketing, growth. One system.

And honestly, it took us a while to get here. We killed our own product twice. Built something, realized it wasn’t enough, and started over. The thing that finally clicked was that AI has to build the tools and run them. Humans steer.

That’s CREAO.

[Evie | WeLike 🦾🍀 @eviegrowth]

After we shared Creao AI’s ambassador program the day before yesterday, the number of sign-ups jumped 5x.

Huge shoutout to @anorth_chen—Abei has been working tirelessly to stay in close touch with everyone.

This ambassador program will cover Chinese, Korean and English markets.

For the English segment, we prefer candidates based in the Bay Area. If you know anyone who might be interested, feel free to refer them! : )

[Evie | WeLike 🦾🍀 @eviegrowth]

Creao AI just opened their Ambassador Program — a rare chance to deeply engage with a fast-growing AI project:

1/ Building AI AgentOS. Their product lets AI run full workflows on behalf of users — from idea to execution in a single flow. Currently in rapid growth mode.

2/ Team has closed 3 rounds, raising tens of millions USD total, backed by top-tier funds like Hillhouse, Sequoia, and Matrix. Founders come from Meta, Alibaba, and other leading tech companies.

3/ This isn't a "just retweet us" kind of ambassador program. Participants get to co-build with the product team, directly connect with the core team, and receive early product access + monthly free credits.

4/ Real economic upside — base fee + 20% commission, plus merch and event reimbursements. Not just promises on a deck.

5/ First cohort only opens slots for Chinese, English, and Korean markets. Limited window — people who know are already moving.

Interested? Apply directly, or DM me for a referral!

[Evie | WeLike 🦾🍀 @eviegrowth]

@KC_PTnDC Feel free to apply!

[Evie | WeLike 🦾🍀 @eviegrowth]

Creao AI just opened their Ambassador Program — a rare chance to deeply engage with a fast-growing AI project:

1/ Building AI AgentOS. Their product lets AI run full workflows on behalf of users — from idea to execution in a single flow. Currently in rapid growth mode.

2/ Team has closed 3 rounds, raising tens of millions USD total, backed by top-tier funds like Hillhouse, Sequoia, and Matrix. Founders come from Meta, Alibaba, and other leading tech companies.

3/ This isn't a "just retweet us" kind of ambassador program. Participants get to co-build with the product team, directly connect with the core team, and receive early product access + monthly free credits.

4/ Real economic upside — base fee + 20% commission, plus merch and event reimbursements. Not just promises on a deck.

5/ First cohort only opens slots for Chinese, English, and Korean markets. Limited window — people who know are already moving.

Interested? Apply directly, or DM me for a referral!

[Creao AI @CreaoAI]

https://t.co/1572tfmVxV

[Evie | WeLike 🦾🍀 @eviegrowth]

https://t.co/hzZ5oA3H5s

[Creao AI @CreaoAI]

Application Form:
https://t.co/O05qDvvS4O

[Evie | WeLike 🦾🍀 @eviegrowth]

After using 20+ Marketing SaaS platforms, my biggest takeaways:

1/ Claude + Skills can essentially replicate customized Marketing SaaS tools. From this angle, most SaaS features — content generation, GTM strategy — are still too thin.

2/ But from another angle, even GTM professionals with technical backgrounds rarely build a “GTM Agent Team” using Claude + Skills, because learning and understanding someone else’s Skills still has a real learning curve.

3/ The newer wave of Marketing SaaS companies are under serious revenue pressure — user onboarding costs are high, individual customers only pay a few thousand dollars, and legacy CRM platforms like HubSpot are also adding AI capabilities.

4/ Many SaaS products pitch the story of automating Creator collaborations, since that business model takes a cut of transactions — which drives more GMV than subscriptions. But in practice, influencer campaigns are still a people business and can’t be fully automated.

5/ There’s still room for SaaS platforms, because so many emerging tech companies lack GTM experience — especially pure-developer founding teams.

6/ To attract these users, beyond data, what matters more is the know-how your product conveys.