Kieran's advantage is building the AI systems he writes about
Most marketing voices report on AI from the outside. Kieran builds the systems inside HubSpot, then teaches what actually worked.
Kieran Flanagan spent a decade running growth and marketing at HubSpot, was CMO at Zapier, and is now SVP of HubSpot's Agentic GTM & Systems group, the team rebuilding go-to-market around AI. He co-hosts Marketing Against the Grain, invests in early-stage AI companies as a Sequoia Scout, and writes a Substack for what he calls the top 5% of marketers. His LinkedIn is not a stream of hot takes about AI. It is a running field report: every week he ships a system, a framework, or a skit drawn from the actual work of turning a marketing org into an AI-native one, and he teaches the lesson in public.
That is the whole engine. Learn-in-public authority is when you become the trusted voice in your field by building real things and narrating what you learn from them, so your own work becomes an audience others can copy. Kieran runs it with unusual credibility: he builds the AI systems at HubSpot, teaches the lesson for free, coins a name for the pattern, and points the depth to a channel he owns.
Reposts every launch as 'INSANE', gates the takeaway behind 'comment for the DM', and leaves you no wiser.
Builds the system himself, teaches what actually worked with the receipts, and hands you a model you can run this week.
“I hope the majority of us choose AI as a craft tool rather than a slop machine.”
— From his post on marketing heading in the wrong direction (691 reactions)
Five findings that repeated across 100 posts
- The receipts are the moat. His most defensible posts show real systems he built, an AI Second Brain, a content agency in Claude Code, with the actual artifacts attached.
- Conversation over virality. He averages 371 reactions but 74 comments a post, a 20% comment-to-reaction ratio, roughly three times the LinkedIn norm.
- Humor carries the reach. His single biggest post is a ChatGPT skit (3,724 reactions), and 'Entertainment' is his second-largest reaction type.
- Every post teaches. New marketing metrics, an agentic org chart, the skills to learn, models a reader can apply that week.
- Weekday discipline. About 2.7 posts a week, Tuesday and Friday heaviest, with just one weekend post across the whole sample.
The numbers behind the account
The story here is not raw reach. It is a steady weekday cadence and an unusually high rate of conversation.
Across the 100 posts we analyzed, Kieran published about 2.7 times a week, almost entirely on weekdays. Tuesday and Friday carry the most volume, and he essentially never posts on the weekend: just one Saturday post and zero on Sunday across the sample. The reach itself is honest and moderate, he averages 371 reactions with a median of 269, and only 4 of the 100 posts cleared 1,000. If you judged him on virality alone you would miss the point. The real signal is in the comments.
When he posts
The content-type mix
Where the engagement comes from
The top posts
| # | Post | Reactions | Comments | Reposts |
|---|---|---|---|---|
| 1 | Me: 'ChatGPT, are these berries poisonous?' | 3,724 | 272 | 60 |
| 2 | The era of AI slop starts with lazy editing | 1,962 | 286 | 44 |
| 3 | I'm officially no longer a marketer | 1,610 | 138 | 32 |
| 4 | A friendly reminder: everyone is overwhelmed by AI | 1,318 | 221 | 40 |
| 5 | Rebuilding your marketing team around AI | 911 | 167 | 22 |
| 6 | We need new metrics for B2B marketing success | 851 | 128 | 27 |
Want to see how your own cadence and comment ratio stack up? Run your profile through our free LinkedIn analyzer.
The six content pillars
Every post is one of six repeatable buckets, so a leader running an AI transformation never runs out of things to teach.
How AI rewrites B2B marketing: new metrics, new skills, new org charts, taught as models you can apply this week.
His own Claude Code builds, an AI Second Brain, a content agency, a Braintrust, shared as blueprints.
The contrarian line that AI should sharpen the craft of marketing, not automate it into slop.
Skits and dialogues about the absurdity of the moment: poisonous berries, comment-gating, AI exhaustion.
Real numbers from rebuilding HubSpot's GTM around AI, the receipts behind the frameworks.
Admitting the overwhelm, the FOMO, and feeling behind, which makes the authority believable.
Pillar 1: The new marketing playbook (the core)
Why it works: He does not just say 'AI changes marketing', he hands over a working dashboard, Influence, Demand, Revenue, that a CMO can present on Monday. Turning a vague shift into a named, copyable framework is what makes a marketing account worth following instead of just scrolling.
Pillar 2: Build-your-own AI systems (the signature)
Why it works: He shows the actual system he built, not a theory about AI. The specificity is the credibility, and it is his most repeated, most defensible pillar because nobody can fake the receipts. Building in public with real artifacts is the thing a competitor cannot copy by paraphrasing.
Pillar 3: Craft vs. slop (the pattern interrupt)
Why it works: Three lines, a coined category ('AI slop'), and 1,962 reactions. His most-shared idea is a stance: use AI to sharpen the craft, never to replace the thinking. A confident line in a feed of AI hype is a pattern interrupt that travels, and it earned 286 comments of agreement and argument.
Pillar 4: Relatable AI humor (the reach engine)
Why it works: His single biggest post is a skit, not a framework. The dialogue lands because every marketer has lived it, and humor is the most shareable format on a feed exhausted by earnest AI takes. It is no accident that 'Entertainment' is his second-largest reaction type.
Pillar 5: Agentic GTM from the inside (the proof)
Why it works: The frameworks land because he has the receipts. Naming real outcomes (345,000 net-new accounts, 82% of chats handled with zero humans) turns an opinion about agentic GTM into proof. Being inside the transformation he writes about is his unfair advantage over every commentator watching from the outside.
Pillar 6: The honest operator (the human)
Why it works: He is the AI authority admitting he is overwhelmed by AI, and that honesty is why the audience trusts the confident posts. Naming the shared feeling ('the AI exhaustion loop') builds more loyalty than another hot take, and it drew 221 comments of 'me too'.
The hooks that earned the comment
His openers are built to provoke a reply, not just a scroll-stop. Most are a skit, a flat correction, or a build you want to see.
Act out the absurd moment as dialogue. 'Me: "ChatGPT, are these berries poisonous?"'
State a hard correction as fact. 'I worry marketing is heading in the wrong direction.'
Lead with what you made. 'I built an AI Second Brain in Claude Code.'
Reframe who you are. 'I'm officially no longer a marketer.'
Promise a fresh playbook. 'We're going to need new metrics to define success in b2b marketing.'
Drop the expert armor. 'As a marketing & growth leader, I've never felt this far behind.'
The through-line is that a Flanagan hook makes a small promise the reader can judge in a second, a laugh, a fight, or a system, then over-delivers on it. For the mechanics of writing openers like these, our guide to writing LinkedIn hooks goes deeper, and you can pressure-test your own first line in the free hook generator.
His top hooks, by the numbers
| Hook type | Opening line | Reactions |
|---|---|---|
| The AI skit | 'Me: "ChatGPT, are these berries poisonous?"' | 3,724 |
| The flat contrarian | 'If you're going to use ChatGPT to write your newspaper articles...' | 1,962 |
| The identity reveal | 'I'm officially no longer a marketer.' | 1,610 |
| The honest reminder | 'Just a friendly reminder, everyone... they are overwhelmed by AI.' | 1,318 |
A voice that reads like a field report, not a think-piece
It sounds like an operator showing you what he built this week, in scannable structure, with the model names and the metrics attached.
- Shows the receipts. Real tools and real model names (Claude Code, VEO3.1, Nano Banana, Opus), and real numbers from inside HubSpot.
- Reframes the news into a lesson. Every launch becomes 'here is what it means for how you market'.
- Writes in scannable structure. Numbered breakdowns, arrow bullets, one idea per line.
- Coins the category. 'AI slop' and 'the AI exhaustion loop' give a feeling a handle that gets repeated.
- Admits the mess. Says out loud when he feels behind or overwhelmed.
- Teaches for free, then points to more. The Substack and the comments carry the depth.
The signature move is turning his own workweek into the curriculum. He is not reporting on AI marketing from the sidelines, he is building the systems at HubSpot and narrating what worked, which is why a post about a Claude Code build reads as a dispatch from the front line rather than a prediction. The credibility is in the specificity: the exact stack, the exact document count, the exact lift.
What he does, and doesn't, do
- Show the actual system he built
- Name the model and the metric
- Reframe every launch as a lesson
- Coin a memorable category
- Admit the overwhelm honestly
- Hype a launch as 'INSANE' with no takeaway
- Gate the value behind 'comment for the DM'
- Outsource the thinking to AI
- Post a vague 'AI changes everything' take
- Pretend he has it all figured out
Holding that voice across frameworks, build logs, skits, and honest confessions at nearly three posts a week, while running an AI org at HubSpot, is the part almost nobody sustains, and it is exactly the gap CaptureFlow closes. 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 voice note, a screen recording of the system you just built, a lesson from a launch), and CaptureFlow, trained on your voice and your past posts, drafts native content for each channel, a LinkedIn post, an X thread, a carousel, a short video, a quote image, so holding the cadence never costs you the craft. See how the AI content agent works.
The systems underneath the posts
Two loops quietly turn 100 posts into an owned audience, authority, and deal flow.
The learn-in-public funnel
LinkedIn is the top of the funnel, not the destination. Kieran gives the gist away free, then routes the depth to channels he owns, so the reach keeps converting into an audience that outlives any single platform's algorithm.
The build-in-public loop
- 1Build a real system for his own workAn AI Second Brain, a content agency, a Braintrust in Claude Code.
- 2Teach the gist on LinkedInThe what and the why, with the receipts, in one skimmable post.
- 3The audience pressure-tests itComments, questions, and his own daily usage surface what to fix.
- 4The system gets betterHe refines it, then builds the next one on top.
- 5The upgrade becomes the next postThe build log never runs dry, because the work is the content.
This builder-educator model is a close cousin of the framework-led one we mapped in the Daniel Priestley playbook, and it is the template most marketing teams should study: do the AI work first, then teach what actually worked, and let the credibility do the selling.
Your 30-day challenge
Run the playbook for a month. Turn your own AI experiments into teaching, then build the loop that compounds them.
- Days 1-2: List every system, experiment, or tool you built or used this month
- Days 3-4: Turn your single best build into a teach-the-gist post with the receipts
- Days 5-7: Reframe one AI launch as 'here is what it means for your marketing'
- Days 8-9: Coin a name for a pattern you keep seeing (your version of 'AI slop')
- Days 10-11: Write one flat contrarian correction of a common AI take
- Days 12-14: Publish a numbered new-rules list for your niche
- Days 15-17: Post one honest line about what is overwhelming you right now
- Days 18-19: Turn an absurd AI moment into a short skit or dialogue
- Days 20-21: Answer a real question your audience keeps emailing you
- Days 22-24: Point one post's depth to a channel you own
- Days 25-27: Rebuild your best-performing post one level deeper
- Days 28-30: Review analytics and double down on the format that reached furthest
Want the cadence without writing every post from scratch while you run the actual work? See pricing to start turning your experiments into weeks of content.
The metrics to track weekly
| Metric | Benchmark to aim for |
|---|---|
| Posts per week | 3+ |
| Comment-to-reaction ratio | 15%+ |
| Comments per post | 50+ |
| Reactions per post | 300+ |
| Named frameworks or systems | 2+ |
| Owned-channel handoffs | 1+ |
The takeaways
- 01Mine your own work for content. Kieran turns the AI systems he builds at HubSpot, a Second Brain, a content agency in Claude Code, into his most defensible posts.
- 02Teach a named framework, do not just react. His biggest teaching posts hand over a copyable model: new marketing metrics, an agentic org chart, the skills to learn.
- 03Coin the category. 'AI slop' and 'the AI exhaustion loop' give a shared feeling a handle, and handles get repeated.
- 04Use humor to beat the fatigue. His single biggest post is a ChatGPT skit, and 'Entertainment' is his second-largest reaction type.
- 05Optimize for the comment. He earns a 20% comment-to-reaction ratio, about three times the LinkedIn norm, by posting takes worth arguing with.
- 06Batch-capture as you build so a nearly-three-a-week cadence survives a heavy launch week.
Frequently asked questions
- How did Kieran Flanagan grow his LinkedIn following?
- By teaching how AI is reshaping B2B marketing, in named frameworks, build-in-public systems, and skits, about 2.7 times a week. Across 100 recent posts he averaged 371 reactions and 74 comments each, and his account passed 112K followers.
- What kind of post performs best for Kieran Flanagan?
- Relatable AI humor and contrarian takes. His top post, a skit about ChatGPT calling poisonous berries edible, earned 3,724 reactions, and an 'AI slop' critique earned 1,962.
- How often does Kieran Flanagan post, and when?
- About 2.7 times a week across the 100 posts we analyzed, almost entirely on weekdays. Tuesday and Friday are his heaviest days, and he posted only once on a Saturday and never on a Sunday.
- How do you apply this playbook without spending hours a week?
- Batch-capture your own experiments and lessons, then let a content agent draft in your voice. CaptureFlow turns one 5-minute capture into a week of native posts across platforms, so you can hold the cadence without writing every post from scratch.
