Felix's unfair advantage is that he ships the thing he posts about
He is not a marketer describing a product. He is the designer who built it, posting the demo the day it goes live.
Felix Haas leads design at Lovable, the AI app builder, and invests on the side. Across the 91 posts we analyzed, published between August 2025 and June 2026, he posted 2.1 times a week and averaged 983 reactions. 60 of those 91 posts mention Lovable by name. The number that stands out is the floor: not one post in the sample finished below 155 reactions. Most accounts this size have a long tail of duds. His does not exist.
The reason is that he is never guessing at the content. Builder-led launch content is when the person who designed the feature posts the demo themselves, so the proof and the announcement arrive in the same object. His biggest post, at 6,386 reactions, is a video of a design feature he spent weeks building, with the line 'It's been my favorite thing to play with for the last few weeks.'
A polished graphic, a feature list, and copy written by someone who did not build it.
A screen recording of the thing working, posted by the person whose hands are in it.
“The next generation of websites will be conversational.”
— Closing line of a post demoing a website he rebuilt as a chat interface (1,917 reactions)
Five findings that repeated across 91 posts
- Video is his strongest format. Video posts average 1,231 reactions against 841 for images, a 46% gap, and video is 36% of his output.
- There is no floor problem. Zero posts finished under 100 reactions, and the weakest in the whole sample earned 155.
- The company is the subject. 60 of 91 posts mention Lovable, and his top six posts are all launches, milestones, or demos.
- His comment average is distorted by a handful of giveaways. The mean is 210 but the median is 67, because a few gated resource posts pulled thousands of comments each.
- He works a six-day week. Only 2 of 91 posts landed on a Saturday, but Sunday carries 12.
The numbers behind the account
A high floor, a strong ceiling, and one metric that needs reading carefully before you copy it.
Felix's median post earns 657 reactions and his mean is 983. 30 of 91 posts cleared 1,000, 10 cleared 2,000, and one reached 6,386. The distribution is unusually healthy: 33 posts in the 100 to 500 band and nothing at all below it. For a 103,000-follower account posting twice a week, that consistency is the headline, not the peak.
The content-type mix
When he posts
The top posts
| # | Post | Reactions | Comments | Reposts |
|---|---|---|---|---|
| 1 | Launching Lovable Aesthetics | 6,386 | 330 | 161 |
| 2 | Announcing our $330M Series B | 4,798 | 263 | 41 |
| 3 | The Lovable x Shopify integration | 3,448 | 322 | 136 |
| 4 | A world where imagination is enough | 3,025 | 175 | 188 |
| 5 | We built our own Brand Agent for Lovable | 2,734 | 310 | 64 |
| 6 | Microsoft now lists Lovable as a requirement | 2,660 | 126 | 68 |
Curious whether your own video posts beat your images? Run your profile through our free LinkedIn analyzer.
The six content pillars
Everything is downstream of the roadmap, which is why the account never has to invent a topic.
A feature ships, and the person who designed it posts a screen recording the same day.
Funding, anniversaries, the new office. Told in first person with the team named.
Someone else's behaviour proving the category is real. Often under 25 words.
Something he made himself, in the product, usually solving his own problem.
Where the design profession is going, with a concrete stack attached.
How he grew the account itself. The most directly copyable thing he publishes.
Pillar 1: The launch video (the reach engine)
Why it works: Fifty-six words over a video. The last line is what a marketer could not have written: 'my favorite thing to play with for the last few weeks' signals he built it, and that single clause is the difference between an announcement and a recommendation.
Pillar 2: The company milestone (the moment)
Why it works: A funding post that spends most of its words on customers and the team rather than the raise. The 500M visits figure does the persuading; the dollar amount is just the headline. Note how low the repost count is at 41: funding news travels as news, not as something people forward.
Pillar 3: External validation (the highest per word)
Why it works: Twenty words, 2,660 reactions. This is the highest reach-per-word post in the sample. He is not claiming the category matters, he is showing that Microsoft's recruiters already believe it does. Third-party proof needs no argument attached.
Pillar 4: The build-in-public demo (the credibility)
Why it works: Twenty-eight words: what he built, what it changes, what it means. He uses his own product on his own website, which is the least deniable form of endorsement available to anyone who works at a software company.
Pillar 5: The designer manifesto (the most saved)
Why it works: The manifesto posts work because they name the tools and the order. A claim that designers can now ship products is arguable; a three-step stack with the actual prompts is usable. 101 reposts, his second-highest, because it functions as a reference.
Pillar 6: The meta playbook (the rarest)
Why it works: He states his own method outright, and the teardown confirms he follows it: document the journey, build the founder brand, post three to five times a week. Worth noting it is one of his lower-reach posts. Audiences reward the work itself more than the explanation of how the work is done.
The hooks that earned the click
Almost every opener names the thing and stops. The flame emoji does the work an adjective would.
Say what shipped, today. 'Today is a big day! We're launching Lovable Aesthetics 🔥'
Four words and a number. 'Announcing our $330M Series B 🚀'
Let someone else make your case. 'Microsoft is now mentioning Lovable as a requirement for their Product Designer roles!'
Lead with the act. 'I rebuilt ChatGPT in Lovable and made it my personal website.'
State the shift as fact. 'Every designer can launch a product in 2026.'
Show the delta. 'This is Lovable's website 1 year ago vs now.'
None of these hold anything back for the second line. He is selling a demo, and the fastest way to get someone to watch a demo is to say what it is. For the mechanics of openers, our guide to writing LinkedIn hooks goes deeper, and you can pressure-test your own first line in the free hook generator.
His line versus the marketing version
- 'Today is a big day! We're launching Lovable Aesthetics 🔥'
- 'Microsoft is now mentioning Lovable as a requirement for their Product Designer roles!'
- 'I rebuilt ChatGPT in Lovable and made it my personal website.'
- 'Every designer can launch a product in 2026.'
- 'We're thrilled to announce an exciting new chapter for design at Lovable.'
- 'The industry is evolving, and enterprise hiring is beginning to reflect it.'
- 'Exploring the future of conversational interfaces: a case study.'
- 'How AI is reshaping the role of the modern product designer.'
A voice that sounds like the person who did the work
Short lines, arrows instead of bullets, and a habit of naming the team rather than the company.
- Keeps it tight. Median post length is 89 words, and several of his highest performers are under 30.
- Uses the flame and rocket emoji as a launch signature, consistently enough that they read as a format marker.
- Names people and teams. The Series B post spends more words on the 120 people than on the raise.
- Uses arrows and unicode bold for structure rather than paragraphs, so a stack or workflow scans instantly.
- Says he built it. Almost every launch post contains a first-person clause placing him inside the work.
- Publishes no plain text at all in this sample. Every one of the 91 posts carries an image or a video.
The launch structure is consistent enough to lift directly, whatever you ship.
[The declaration: 'Today is a big day! We're launching ___ 🔥'] [What it now does, in one or two plain sentences, no feature list] [The builder's clause: 'It's been my favorite thing to play with for the last few weeks.'] [The invitation: 'Go check it out!'] [A video of the thing actually working]
What he does, and doesn't, do
- Post the demo the day it ships
- Put himself inside the work, in first person
- Let third parties make the category argument
- Name the team and the customers
- Attach a video whenever there is something to show
- Plain-text posts entirely
- Feature lists and spec sheets
- Corporate announcement language
- Long posts; his median is 89 words
- Explaining the significance instead of showing it
The constraint on this model is obvious: it only works while there is a roadmap shipping behind you. Twice a week, every week, with a video attached, is a production load that most product teams cannot carry alongside the actual building. CaptureFlow is an AI content agent that turns your expertise into weeks of on-brand content for every platform. You capture one idea in 5 minutes, a screen recording of the feature, a voice note on why you built it that way, and CaptureFlow, trained on your voice and your past posts, reshapes it into a LinkedIn post, a short video, a carousel, and an X thread. See how the AI content agent works.
The system underneath the posts
The roadmap is the content calendar, and three people at the same company are running it in parallel.
The ship-to-post loop
- 1A feature shipsAesthetics, Shopify, General Tasks, Themes, the new dashboard.
- 2The designer records it workingVideo posts average 1,231 reactions against 841 for images.
- 3The post carries a builder's clause'It's been my favorite thing to play with for the last few weeks.'
- 4Users try it and build with itTheir apps generated 500M+ visits in six months.
- 5Their results become the next milestone postAnd the loop restarts one tier bigger.
The Lovable cluster
This is the third Lovable account we have taken apart, and the three are running visibly different jobs. Anton Osika, the CEO, narrates the company's milestones. Elena Verna, who leads growth, publishes the growth thinking. Felix ships the product proof. Same company, three audiences, almost no overlap in what they post.
Milestones, hires, offices, and the mission. The company as a story.
The growth thesis, written for operators who want the mechanism.
The demo, the day it ships, recorded by the person who designed it.
Read this one alongside the Anton Osika playbook and the Elena Verna playbook to see the full picture. For anyone building in public, the three together are the clearest template we have for splitting a company's story across more than one voice.
Your 30-day challenge
Nine posts. Every one of them showing something that now exists and did not before.
- Days 1-2: List everything your team shipped in the last month that nobody announced
- Days 3-4: Record a screen video of the best one and post it the same day
- Days 5-7: Add the builder's clause: one sentence placing you inside the work
- Days 8-10: Find one piece of third-party evidence your category is real and post it in under 30 words
- Days 11-12: Use your own product on something personal and show the result
- Days 13-14: Compare your video reach against your image reach honestly
- Days 15-17: Write the manifesto for where your profession is going, with the actual stack
- Days 18-19: Name the team, not the company, in your next milestone post
- Days 20-21: Cut every feature list from your drafts
- Days 22-24: Decide which beat you own and which a colleague should own instead
- Days 25-27: Get a second person at your company posting on a different beat
- Days 28-30: Check your worst post of the month, and raise the floor
The stop-doing list
| Stop doing | Do this instead |
|---|---|
| Letting marketing announce what you built | Posting the demo yourself, the day it ships |
| Screenshots of software | A video of it running |
| Feature lists | One sentence on what it now does |
| Claiming the category matters | Showing that someone else already believes it |
| One executive carrying the whole company | Three people on three distinct beats |
| Chasing a peak | Raising the floor; his worst post in 91 earned 155 |
The metrics to track weekly
| Metric | Benchmark to aim for |
|---|---|
| Worst post of the month | Above 100 reactions; his floor is 155 |
| Video share of output | 35%+, if you ship software |
| Ratio of video reach to image reach | Measure it before you assume |
| Median post length | Under 100 words |
| Posts within 24 hours of a ship | Every launch |
| Colleagues posting on a different beat | At least one |
The takeaways
- 01Post the demo yourself. Felix Haas's biggest post, at 6,386 reactions, is a video of a feature he designed, published the day it shipped.
- 02Video beats images for software. His video posts average 1,231 reactions against 841 for images, a 46% gap.
- 03Raise the floor. Not one of his 91 posts finished below 155 reactions.
- 04Let third parties argue your case. Twenty words about Microsoft listing Lovable in job requirements earned 2,660 reactions.
- 05Add the builder's clause. One first-person sentence placing you inside the work turns an announcement into a recommendation.
- 06Split the beats. Lovable runs three accounts on three distinct subjects rather than one executive covering everything.
Frequently asked questions
- How did Felix Haas grow his LinkedIn following?
- By posting product demos the day they ship, as the designer who built them. Across 91 recent posts he averaged 983 reactions on 2.1 posts a week, with 60 of those posts mentioning Lovable and not one finishing below 155 reactions.
- What kind of post performs best for Felix Haas?
- Launch videos. His top post, launching Lovable Aesthetics, earned 6,386 reactions, and his video posts average 1,231 reactions against 841 for image posts.
- How many Lovable people has CaptureFlow analyzed?
- Three. Anton Osika, the CEO, narrates company milestones; Elena Verna, who leads growth, publishes the growth thinking; Felix Haas ships the product proof. Same company, three separate beats, very little overlap.
- How do you keep a demo-led cadence when the roadmap is quiet?
- Batch-capture the small things as they happen: a clip of a fix, a note on a design decision. CaptureFlow turns one 5-minute capture into a week of native posts and short video across platforms, so a slow sprint still produces real content.
