How do you build in public without spending hours? You stop treating writing as the work. Capture ideas as voice notes during dead time. Let a pipeline trained on your voice turn them into drafts. Keep one job for yourself: a one-tap yes or no on every post. My version of this costs me about two hours a week. It produced 4 million LinkedIn impressions in under a year, in a topic where I started with zero expertise.
Building in public fails on time, not talent. The fix is a capture-first system: ~2 minutes per idea by voice, drafting done by AI in your extracted voice, ~10 minutes per approval sitting, 10-15 minutes a day of commenting. Below: my real weekly budget, my real launch funnel (150 visitors, 21 signups, 5-8 activated), and the CTA placement mistake that costs founders 10x.
The lie in "building in public takes hours"
It does take hours. If you write at a desk.
Desk writing is the default advice, and it is why most founders quit by week three. Shipping always beats posting when both compete for the same hours. So they must not compete.
My best ideas never show up at my desk. They show up walking, driving, after a customer call. For two years I lost them, because "write it later" is where ideas go to die. The system below exists to catch them in the moment, at a cost of about two minutes each.
The weekly budget
This is my actual week, not an aspiration. Times are approximate and honest.
| Activity | Time | When it happens |
|---|---|---|
| Capturing ideas (voice notes) | ~20 min total, 2 min each | Dead time: walks, driving, between calls |
| Approving, tweaking, or skipping drafts | ~10 min, twice a week | Phone, usually evening |
| Commenting on others' posts | 10-15 min a day | Coffee, queue, commute |
| Writing something fully by hand | ~0-30 min, some weeks | Only when a post deserves it |
Total: roughly two hours a week of attention. Almost none of it at a desk.
Notice what is missing: brainstorming sessions, content calendars, blank pages. The raw material is my actual week. A customer call becomes a post about what the customer taught me. A bug becomes a post about judgment. The commute becomes the capture window.
The system, step by step
- Capture by voice. I talk into Telegram while walking. In Hebrew, usually, mid-thought, unpolished. The recording gets transcribed automatically and lands in my content pipeline as a "seed". I have accumulated 180 of these. Each cost me about two minutes.
- Draft in my voice. The pipeline (I use Liftli, which I built, inside Claude) matches each seed to my content strategy and drafts a post in my extracted voice. Not "AI voice". Mine, learned from my own writing. Each draft passes a critique-and-revise loop before I ever see it.
- The one-tap gate. Nothing publishes without me. I approve, tweak, or skip from my phone. This is the whole quality mechanism, and it takes minutes because I am editing my own ideas, not generating from nothing.
- Comment daily, briefly. Commenting is the most underpriced activity on LinkedIn. It builds relationships and reach at a fraction of the effort of posting. My comments have been recognized in sales calls before I ever introduced myself.
What a system produces (when talent is not the input)
A year before this, I had zero expertise in the topic I started writing about. No audience for it, no playbook.
The system produced 4 million impressions in under a year. Along the way I led a community of 10 authors that reached roughly 100 million cumulative impressions. I am not a naturally gifted writer. That is the point. Consistent capture plus consistent publishing plus consistent commenting beat talent, because talent does not show up every week and a system does.
The launch data nobody posts
Building in public means showing the math on the ugly days too. Here is the first week of my SaaS launch on LinkedIn, unedited:
| Launch funnel, first 3 days | Number | Rate |
|---|---|---|
| Website visitors | ~150 | — |
| From my profile CTA | ~70 | ~50% of traffic |
| Completed signups | 21 | 14% of visitors |
| Product activations | 5-8 | — |
| Signups on mobile | ~60% | couldn't install the desktop extension |
My launch post reached fewer than 200 people in its first hour, despite roughly 1-in-8 of them engaging. Then LinkedIn showed me the boost button, like a rainbow after rain. I took a breath and did not press it.
Two lessons from that week that I have never seen in a growth playbook:
- CTA placement is a 10x variable. A direct link at the end of a short post converted 1 in 25 readers. The "visit my website" button on my profile converted 1 in 150-200. Same audience, same week. And linking to a LinkedIn company page instead of your site doubles the friction: one more scroll, one more click.
- Sixty percent of my signups came from mobile and then hit a desktop-only install step. Whatever you launch, walk the mobile path first.
The part fear plays
I delayed my launch by two months. Not for engineering reasons. I can name the five fears exactly: the idea might be stupid, the product was not good enough, people would say "this is what he's doing now?", the AI backlash, and the possibility of zero traction forcing me to quit in 30 days.
The reframe that unstuck me: the asset I am building is not the startup. It is the founder brand, and it survives the startup. Every post compounds into something that moves with me, whatever happens to the product. Once I saw it that way, a slow launch day stopped being a verdict.
Steal the system
You do not need my tool to run this loop. You need the four parts: frictionless capture, drafting that sounds like you, a human gate, and daily small-scale commenting. Start with voice notes and a transcription app if you want to wire it yourself.
If you want the loop pre-built inside the AI you already use, that is what Liftli is. It is the system in this post, productized: your voice notes become drafts in your voice, and nothing ships without your tap.