Dharmesh wins with ideas, not production value
He is the co-founder and CTO of a public company with a million-plus followers, and 70% of what he posts is unadorned text. The reach comes from the thinking, not the format.
Dharmesh Shah co-founded HubSpot in 2006, took it public, and is still its CTO. He also builds constantly on the side: Agent.ai (2.5M users), HubCode, HubGrader, a simple.ai newsletter with 2M+ subscribers. His LinkedIn is where all of that thinking gets aired first, usually as a plain-text post typed in the same terminal-brained voice he codes in. We read 100 of his most recent posts to map how a builder, not a broadcaster, grew past 1.2M followers.
His whole method has a name he would probably resist, but it fits. Building in public is narrating what you are making and learning in real time, so your audience becomes your first users and your feedback loop. Dharmesh does not wait for a launch to be polished. He posts the tinkering, asks what people think, and ships based on the replies.
Waits for the finished launch, posts the press-release version, and treats the audience as a megaphone.
Posts the half-built idea, asks 'is this a good idea?', and lets the comments shape what he builds next.
Plain text, outsized reach
The numbers say something counterintuitive: the guy with the biggest audience mostly writes text, and it works.
Across the 100 posts we analyzed, Dharmesh averages 514 reactions a post and clears 1,000 reactions 17 times. His comment-to-reaction ratio is 18.9%, roughly three times the typical LinkedIn norm. But the real story is the format mix.
Read that twice. He could clearly lift every post by adding a graphic or a video, and he mostly does not. He optimizes for volume and speed of thought over polish, because his edge is the thinking. Curious how your own format mix and reach compare? Run your profile through our free linkedin analyzer.
| # | Post | Reactions | Comments | Reposts |
|---|---|---|---|---|
| 1 | Woo hoo! HubSpot turns 20 years old today! | 4,036 | 312 | 102 |
| 2 | HubSpot co-founder/CTO buys $1.8M of his own shares | 3,490 | 314 | 61 |
| 3 | The SCARIEST and/or CRAZIEST thing I've done in years | 2,291 | 284 | 69 |
| 4 | BREAKING NEWS: Andrej Karpathy joins Anthropic | 2,243 | 159 | 40 |
| 5 | Tinkering on a new idea... Hub Code | 2,095 | 420 | 33 |
| 6 | Sneak peek of HubCode, vibe coding for HubSpot | 2,086 | 196 | 82 |
Six pillars, from news bulletin to human
His feed looks like a stream of consciousness and is actually six repeating pillars, each with its own job.
An AI or industry development, plus the insider take only he can give.
Current Status updates on HubCode, Agent.ai, and HubGrader.
Where B2B software is heading: agents that run your product, not just use it.
Launches, anniversaries, and unabashed love for the company.
Mantras and lessons from a 30-year career, kept short.
Piano streaks, chess losses, and proof he is not a bot.
1. BREAKING NEWS curation
Why it works: Why it works: he wraps a dense AI story in a news-bulletin frame, then adds a take with the credibility of someone who ships AI daily.
2. Build in public
Why it works: Why it works: he shows the thing he built, names the exact problem it solves, and invites feedback. The build is the content.
3. The agentic thesis
Why it works: Why it works: a clear, quotable thesis that names the shift before most people see it. Prediction posts get saved and shared.
4. HubSpot milestones
Why it works: Why it works: a bittersweet hook ('my last') on a routine rebrand announcement, so a company update reads like a personal moment.
5. Founder wisdom
Why it works: Why it works: a short, memorable maxim earned over decades. No thread, no framework, just one line people can carry.
6. The human behind the builder
Why it works: Why it works: a deliberate break from the AI firehose. The chess losses and piano streaks make the authority feel like a person.
Three hooks that make dense ideas feel light
Dharmesh writes about heavy technical topics, and yet the openers feel breezy. That is by design. He uses a small set of familiar frames.
BREAKING NEWS: [an AI or industry development] ... then his insider take Woo hoo! [a HubSpot or personal launch worth celebrating] Current Status: [what he is tinkering on right now] The trick: a familiar frame (a news bulletin, a cheer, a status line) that makes a dense idea feel like a bite, not a lecture.
The frames repeat because they work. Each one sets an expectation the reader can process in a second:
- The news bulletin: 'BREAKING NEWS: Andrej Karpathy Joins Anthropic.' He curates the moment, then adds the take only he can.
- The cheer: 'Woo hoo! HubSpot turns 20 years old today!' Genuine, uncynical enthusiasm as a hook.
- The status update: 'CURRENT STATUS: Tinkering on a new idea... Hub Code.' An open invitation to watch him build.
- The confession: 'This might be the SCARIEST and/or CRAZIEST thing I've done in years.'
- The one-word cold open: 'Goosebumps.' One word, then the story.
The through-line is that a hook is a promise about how much effort the post will cost the reader. Want to draft your own in this style? Try our free hook generator, and our guide on how to write LinkedIn hooks breaks the patterns down further.
The build-in-public flywheel
The pillars and hooks feed one loop that turns Dharmesh's tinkering into an audience and that audience into users for the things he builds.
- 1Tinker in publicPosts Current Status and sneak-peek updates while building HubCode, Agent.ai, and HubGrader.
- 2Ask for feedback'Feedback is a gift.' A like or a comment shapes what he builds next.
- 3Launch to the same crowdThe people who watched him build become the first users on day one.
- 4Repurpose one idea everywhereA single write-up becomes an avatar video, a long-form post, and a short update.
That last step is the tell. Dharmesh literally built an agent to turn his written posts into avatar videos, because, in his words, 'I write a lot of content, but most of it reaches only a subset of its potential audience.' That instinct is pure capture-first: write the idea once, then reshape it for every audience. It is exactly what CaptureFlow does. CaptureFlow is an AI content agent that turns your expertise into weeks of on-brand content for every platform, trained on your voice, so one captured idea becomes a LinkedIn post, an X thread, a short video, and a carousel without you rebuilding it five times.
The humble builder: his voice rules
For a founder of his stature, the voice is strikingly low-ego: curious, generous, and allergic to hype.
- Writes in plain, conversational, terminal-brained text
- Shares the messy work-in-progress, not just the launch
- Credits people by name and calls feedback a gift
- Adds a genuine 'Woo hoo!' or a self-deprecating aside
- Explains the technical idea for 'mere mortals'
- Leans on polish or production to carry a weak idea
- Postures as the smartest person in the room
- Hypes without a caveat or a disclosure
- Hides the failures (the lost chess games stay in)
- Talks down to a reader who is newer than he is
It is a voice built for the founder building in public, the same lane Anton Osika runs in our Lovable teardown. The lesson both prove: sharing the build, humbly and often, beats announcing the finished thing.
Steal Dharmesh's build-in-public playbook in 30 days
You do not need a public company or a million followers. You need something you are building or learning, and the willingness to narrate it.
- Pick one thing you are building or learning right now
- Post a 'Current Status' update on it, half-finished is fine
- End with a real question and ask for feedback
- Find one development in your field this week
- Post it as 'BREAKING NEWS' plus the take only you can give
- Keep it plain text. Speed and clarity over polish.
- Write one short mantra from your own experience
- No thread, no framework, just one line worth carrying
- Add one human post so you are not only your work
- Take your best post and reshape it into a video or carousel
- Share the same idea where a new audience will see it
- Turn the people who engaged into a launch list
Dharmesh does the repurposing by hand, and even coded his own agent for it. If you want the same one-idea-everywhere output without building the tooling yourself, that is what CaptureFlow automates. See pricing to start turning your own work-in-progress into weeks of content.
The takeaways
- 01Dharmesh Shah, HubSpot's co-founder and CTO, grew past 1.2M LinkedIn followers while posting mostly plain text, not video or graphics.
- 0270% of his posts are plain text, yet his rare videos and images earn the most reactions. The ideas carry the reach.
- 03His signature format is 'BREAKING NEWS': he curates an AI or industry development, then adds the insider take only he can give.
- 04He builds in public relentlessly, posting 'Current Status' updates on tools like HubCode and Agent.ai and asking his audience for feedback.
- 05He repurposes one idea across formats, even building his own agent to turn written posts into avatar videos for people who prefer video.
- 06His voice is humble and optimistic: 'Woo hoo!', 'Feedback is a gift', and a mantra of 'Dream big, iterate small.'
Frequently asked questions
- Who is Dharmesh Shah?
- Dharmesh Shah is the co-founder and CTO of HubSpot and the creator of Agent.ai. He has grown to over 1.2M LinkedIn followers by sharing his AI experiments, his build-in-public projects, and his take on where B2B software is heading.
- Why is Dharmesh Shah's LinkedIn so effective?
- He mixes three things: 'BREAKING NEWS' curation of AI developments with an insider take, build-in-public updates on his own tools, and genuine, humble enthusiasm. Across 100 recent posts he averages 514 reactions, and 70% are plain text.
- What is Dharmesh Shah's content strategy?
- Write the idea once, then reshape it for every audience. He posts mostly text on LinkedIn, runs a 2M+ subscriber newsletter, and even built an agent to turn posts into videos, all from the same core ideas.
- How many followers does Dharmesh Shah have on LinkedIn?
- Over 1.2 million. He built that audience over years by consistently sharing what he is learning and building in AI and B2B software, mostly in plain-text posts.
