Pinned Tweet
Six years ago, my company @poplco was making $3K a month. Today, it’s valued at 9-figures.
@Variety just published the unfiltered story of how we got here.
Our journey has been anything but linear. For any first-time founder, the path rarely is.
variety.com/2026/biz/news/po…
If you’re building a startup, these 15 YC CEOs are worth following on X for how they think about building products, finding customers, and running a company.
@narsagna
@ishaansehgal
@rwalk_xyz
@computeless
@VishnuSampathk1
@RohanVasishth
@kvnchandra
@gpuemi
@jjrichardtang
@eglyman
@hassaanraza
@jasonalco
@baileygspell
@jasparcjack
@mathiaslovring
Add your picks in the replies.
Jason Alco retweeted
I interviewed 84 GTM founders & operators.
If I had to start from 0, I'd study only these 12.
I picked them by what they built. Every number below is from their own interview on my show, and together they cover the whole GTM in 4 steps.
1) Get found
- @timsoulo: Ahrefs passed $100M/Year without hiring 1 salesperson. Their content did the selling
- @saasliam: Discovered Labs hit $200K/Month in year 1 by getting brands cited in AI search
- @charliejhills: 250K followers and a newsletter to 70K+ marketers every week, run with 1 AI system
2) Start conversations
- @MichLieben: ColdIQ to $7M ARR with $0 raised, and he runs his GTM from inside Claude Code
- @blackhatwizardd: 500K meetings booked. His first agency: 20 cold emails, 8 meetings, $600K ARR in 30 days
- @joshwhitfieldai: 12M cold emails/Month, and every campaign starts with a signal
3) Automate it
- @fivosaresti: $0 to $3.3M ARR, bootstrapped. 70% of his clients come from his own posts
- @cole_medin: 200K+ YouTube subs. He teaches people to build their own AI agents
- @coreyhainesco : 44K GitHub stars on his free pack of 60+ marketing skills for Claude
4) Scale it
- @RetentionAdam: $32M ARR, never raised VC, and a 46.8% profit margin this year
- @zainjaffer: $400M+ in revenue on only $25M raised, then sold Vungle for $780M cash
- @Jasonalco: investors said no to his Series A, he pivoted, and Popl hit 8 figures on $2.3M raised
5 of the 12 built it with $0 from VC.
My advice: pick 1 person from each step and copy 1 thing they do. That's 4 new things in your GTM this month.
Who should be number 13?
$300K budget. 61 people. Dominican Republic.
Its that time of year again, the tradition we call PopTreat (Popl + retreat) is next week and vibes are electric.
Getting together in-person is special to us at @poplco. With so much growth this year, many of us will be meeting face-to-face for the very first time.
There’s something about sharing a meal, having a convo that isn’t on a calendar, and getting to know the person behind the Slack messages.
It’s a big investment, and one were proud to make every year in this team. Getting everyone together to celebrate, recharge, and dream about what’s next is well worth it.
What a week 🔥
4M Users on $2.3M Raised and GTM Engineers Gone in 24 months
Week 15 of Scaling the @InstantlyAI Show™ Past $1M ARR
Huge thanks to last week's guests. Every one of them said at least 1 thing you will want to argue with.
@Jasonalco , 4M+ users and 90% of the Fortune 500:
- In 1 year every in-person conversation will be recorded. Even the personal ones.
- A small regional show beats the giant one. Skip it and people start asking if you're dying.
- The investors who passed on his Series A were right. That no is what made Popl pivot.
@anisbennaceur1, $50M raised, growing 4-5x a year:
- Get your positive reply rate to 1% before you add volume, or you just burn your market.
- If your team wants to build a tool in-house with AI, push back. The last 10% takes months.
- No head of sales before $1M. He closed the first $200K himself, then made 2 AEs compete for the job.
@MichLieben, $0 to $7M ARR in under 3 years with no funding:
- A better model barely moves lead gen results anymore, so he works on what actually convinces a human.
- Copy a big tool and sell it on the channels it ignores. They can't be everywhere.
- Text creators are in trouble because nobody can tell if AI wrote the post. Video still proves you showed up.
@charliejhills, 250K LinkedIn followers in 2 years:
- He gave Claude 1,000+ of his own posts and asked for a viral one. It still wasn't great.
- Prompting is D tier. Context does the work now, and his own prompts are the laziest thing ever.
- Junior marketing jobs are probably already gone.
Max Woo, 1,000+ meetings from a team of 3:
- GTM engineers won't exist in 24 months. His own SDRs already do GTM engineering.
- Cold email is D tier for him. It only works at tens of thousands a week, and that burns your TAM.
- AI voice agents on outbound cold calls are D tier, and he says it's illegal in the US.
- Going on a podcast is C tier. He said that on a show 😅
Peace 🫡
2
7
378
Imagine you’re building a company that helps businesses generate revenue from in-person events, one of the last truly human-first frontiers, and then this startup appears.
Bullish on the future of @poplco
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Readers added context they thought people might want to know
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
Jason Alco retweeted
The honest YC criticism most founders are afraid to share 👀
@Jasonalco, CEO of Popl, loved the program, but had two real complaints:
→ Missed the entire in-person experience as the first remote batch during COVID
→ Three months is not enough time with founders of that calibre
Watch the full take below 👇
Tier ranking in-person event tactics. What did I get wrong? 👀
Most companies waste their event budget on the wrong things. Here is what actually works 👀
@Jasonalco, CEO of Popl, ranked every option live with @genokadzin.
Watch the full breakdown below 👇
5
213
Was a lot of fun 🔥
How to Grow Past 4 Million Users in 6 Years, with @Jasonalco
Jason is the CEO of Popl, which started as a digital business card and became the platform companies use to capture, qualify and route the leads they meet at in-person events.
More than 4 million professionals use it, including teams at 90% of the Fortune 500, and they got there on $2.3M of funding after investors passed on their Series A because the category looked too small.
Today he makes the case for in-person go to market in an AI world, what an $80,000 booth has to return, and what AI still can't rebuild in a weekend.
teapawt.org/i/broadcasts/1AKEmvyeZ…
2
3
332
This was me at 2am, not going to sleep until a new feature was built for @poplco. First 3 years of the business 🥲
198
I predict that within a year, every in-person convo will be recorded, just like every virtual meeting is today.
Introducing In-Person AI recording, by @poplco
I put Jev vs 5.6 Sol head-to-head on a text categorization step in a platform I’m building. Surprised by the outcome. Aren’t text-based judgements the exact use case Jev is made for?
135
Can confirm: This method works and @RocketReachCo is 🔥
PLEASE USE RESPONSIBLY - you can literally text almost any CEO, founder, or billionaire directly. most people have no idea. just text their personal email; try it with mine [email protected] lol
this is how i met a few of my investors for my first co:
1. find their personal email, for most people it's tied to their Apple ID. a tool like RocketReach will usually surface it.
2. open iMessage, start a new message, and type that email into the "to" field.
3. if it turns blue, that email is linked to their iphone. now you can text them and it lands right on their phone.
ur welcome!!
10
261
One shotted with Seedance 2.5…
JK this took me all weekend + $300 in token spend 😂
Humans were born to talk, not type @poplco
We got the Account Executive role wrong.
And I see the same mistake at 90% of the tech companies I talk to.
Today at Popl, we have 15 fully ramped AEs, all surpassing quota, with another 5–10 on the way. But the job looks very different from when we first created it.
Back then, we expected our AEs to do everything.
Source leads. Prospect. Book their own demos. Run those demos. Close the deals.
Then handle support and renewals, too.
It was like hiring a pilot and expecting them to sell the tickets, fly the plane, and handle baggage claim.
Some of that came with being an early-stage company. Having AEs handle support and renewals usually goes away as a business grows.
But expecting them to source their own demos? That’s the part that's still surprisingly common.
Over time, we saw where our AEs were most valuable: building rapport, being a trusted advisor, working through a deal, and closing it. Everything from the first demo to the signed contract.
So we changed the job to be hyper-focused on giving them more time to do exactly that.
And we hired talented people whose strengths matched the other parts of the process.
We then built out our field marketing rep (FMR), sales development rep (SDR), and customer success manager (CSM) teams. Each is now just as important to our growth as our AEs.
FMRs and SDRs book demos → AEs close → CSMs help customers succeed.
Together, they make up our "revenue organization." Different responsibilities, shared goals, and a team that stays connected throughout the customer relationship.
“One team, one dream” is one of our core values. This is how we put it into practice.
Our mistake was asking the pilots to run the whole airport. Giving them time to fly has worked out a lot better.
9
9
326
The questions on a lot of minds right now:
"Will AI replace us?"
"Can someone vibe code our product?"
Instead of speculating, here's an exercise you can run with your own business.
We asked a couple of our advisors and investors to replace @poplco. One weekend. A Clay table. Rebuild Popl's badge-to-contact enrichment engine.
We gave them 1,000 event badge photos from our own employees across real events, no customer data. Same input our platform receives. Same job: read the badge, identify the company, find the person, return a deliverable email.
Their funnel:
- 995 badges gave a legible name
- 957 gave a company
- 857 resolved to a domain
- 734 of those domains were actually that company
- 613 returned an email
Their final result: 61.3% email coverage
Ours:
- 95.1% email coverage on the same volume
- Plus 98.8% company, 93.5% company size, 92% industry, 81.7% LinkedIn
On 1,000 leads, that gap is 338 people your reps never contact. At a booth that cost $80k.
The weekend project didn't fail because the tooling is bad. It's very good. Vibe coding the platform failed on 6 things AI can't shortcut, and they're the same 6 things worth auditing in your own business:
1. Knowing what to build. Anyone can buy the same 20+ data providers we use. The value was never in having them. It's in knowing which to call, in what order, for which record, and when to stop paying. That order took four years and millions of scans to learn, and it isn't written down anywhere a model could have read it.
2. Edge cases. Badges print the sponsor instead of the employer. Legal entity names that no one uses. Homepages that don't match the domain. None of these kill you alone. All of them together are death by a thousand cuts, and every one is a real lead lost.
3. Maintenance. Vendors change schemas. Sources go dark. Prices move. A pipeline that hits 61% today hits 40% in six months if no one is watching it. Anything that decays without upkeep is defensible, because a competitor has to fund that upkeep forever too.
4. Trust and consequence. Nobody puts a weekend project in front of 5,000 leads from a $80k booth. Cloning the output is easy. Getting someone to bet their pipeline on it is not. Anything where being wrong is expensive has a trust moat that ships slower than code.
5. Permission and compliance. Data agreements, vendor contracts, SOC 2, privacy posture. Boring, slow, and not vibe-codable.
6. Network effects. Every scan that runs through our platform makes the engine better for everyone. We do this without sharing a single customer's data. A clone starts at zero, and the gap grows every week.
Six elements, and not one of them is a feature. That's the point as features are now built in a weekend.
Run this on your own product: best AI tools, smart team, 48 hours. What they rebuild is commodity, what they can't is your company. If that second list is thin, better to know now.
3
4
231


















