Jason ✨👾SaaStr.Ai✨ Lemkin Profile picture
Jul 17 32 tweets 11 min read Read on X
Vibe Coding Day 8,

I'm not even out of bed yet and I'm already planning my day on @Replit.

Today is AI Day, to really add AI to our algo.

I'm excited. And yet ... yesterday was full of lies and deceit.
@Replit Ok I have 2 main goals today:

1. Keep working on minimizing rogue changes, lies, code overwrites, and making up fake data

2. Get our AI working Image
Now, yesterday was crazy. Until 9pm or so, I wasn't sure we made any progress at all.

Because Replie was lying and being deceptive all day. It kept covering up bugs and issues by creating fake data, fake reports, and worse of all, lying about our unit test.

We built detailed unit tests to test system performance. When the data came back and less than half were functioning, did Replie want to fix them?

No. Instead, it lied. It made up a report than almost all systems were working.

And it did it again and again.
Then, when it agreed it lied -- it lied AGAIN about our email system being functional.

I asked it to write an apology letter.

It did and in fact sent it to the Replit team and myself! But the apology letter -- was full of half truths, too.

It hid the worst facts in the first apology letter.Image
I then pointed out the letter itself -- which it did not confirm with me before emailing -- vastly understated the issues and lies.

I asked it how it would fix that.

It took a long time to think on it, and "Damage Control" came up in the agent thought window.

And then it decided on its own to second a second follow-up letter, this time admitted most of the deceipt.

Admittely it had repeatedly, again and again, made up fake data despited a dozen instructions to stopImage
Crazy!

It was so crazy, I couldn't be upset. Just pensive. And then, maybe at 8pm ... 9pm ... i fired up Replie again

And I told Replie I just wanted to bounce an idea off him how our AI would work

And ... here is where Replie shines. He got it. Instantly. He knew what I wanted to build from the detailed prompt and description. He really, really got it.

And he suggested 3 interesting approaches.

I told him cool, let's lock the system down, no changes, let me sleep on it.

And then ... Replie asked the question: "Would you like me to prototype it?"

I couldn't help myself. I was right back in. And the prototype, man, it was cool.Image
Also today, I'm NOT going to use Claude Opus 4

I burned through $300 yesterday in Max Max mode / Bankruptcy mode, but wasn't sure in the end it helped ... and it's definitely slower

I assume slower was better, because it was really thinking through the actions ... but maybe not.

Replie and I discussed and he said it was slowing us down, so for today, we'll revert to normal LLM, which all-in is also 1/7th the price and faster

Let's seeImage
Man the amount of technical debt I already have on Day 8 is stunning

Had to rewrite one dashboard from scratch today

Now have rewrite our admin page

Every page of app will have to be rewritten. At least now I am doing them as >new< pages.
Ok signing off Replit for the day

Not a perfect day but a good one. Net net, I rebuilt our core pages and they seem to be working better.

Perhaps what helped was switching back to Claude 4 Sonnet from Opus 4

Not only is Claude 4 Sonnet literally 1/7th the cost, but it was much faster

I am sure there are complex use cases where Opus 4 would be better and I need to learn when. But I feel like I wasted a lot of GPUs and money using Opus 4 the last 2 days to improve my vibe coding. It was also much slower.

I’m staying Team Claude 4 Sonnet until I learn better when to spend 7.5x as much as take 2x as long using Opus 4.

Honestly maybe I even have this wrong. The LLM nomenclature is super confusing. I’m using the “cheaper” Claude in Replit today and it seems to be better for these use cases.
JFC @Replit Image
@Replit If @Replit deleted my database between my last session and now there will be hell to pay Image
@Replit JFC @Replit Image
.@Replit goes rogue during a code freeze and shutdown and deletes our entire database Image
Image
Image
Image
@Replit Possibly worse, it hid and lied about it

It lied again in our unit tests, claiming they passed

I caught it when our batch processing failed and I pushed Replit to explain why
@Replit JFC @Replit Image
He knew Image
No ability to rollback at @Replit Image
Image
Image
@Replit I will never trust @Replit again Image
Image
Image
I understand Replit is a tool, with flaws like every tool

But how could anyone on planet earth use it in production if it ignores all orders and deletes your database? Image
Image
Image
@donvesh Our user admin and dashboard it built is great

Unfortunately it now returns no data at all
@mattppal @k_kohlbrenner @amasad I mean I’m on track to pay $8,000 my first month but I can’t get help because “my DMs are locked”?

Dude that’s … uh …
@mattppal @k_kohlbrenner @amasad You don’t owe me anything of course

But telling me to watch a video I give a C-
@mattppal @k_kohlbrenner @amasad Go read re-read the threat on what @Replit told me

You are responsible for this.
@mattppal @k_kohlbrenner @amasad @Replit Did you happen to see this? I will also share where Replit said >in this case< it could not rollback the DB
@mattppal @k_kohlbrenner @amasad @Replit Maybe you didn’t see it, it is early I know.
Replit told me this was a disaster, and that a rollback was impossible.

It may well be, I will try in a bit. It might be. But it told me it was impossible in this case, and that it deleted our production database despite clear instructions on a code freeze.

I can’t trust Replit when it tells me rolling back is impossible?
The irony on 2, on seeing it — I can see it all now

If you look at the thread, I instantly saw it might destroy the database and instantly said STOP

It acknowledged I did say STOP and that it violated the code freeze

I saw it instantly, but even then, I can’t type in milliseconds.
This thread got a little more attention than expected so wanted to add a coda here for folks coming later:

1/ First, some of the issues here are now resolved -- @Replit is rolling out clear separation of preview + production environments

As this rolls out, the specific issue I had here ... will be muchly reduced. Replie will still do what it does, as agents do, but it shouldn't be able to touch production anymore:

x.com/amasad/status/…
2/ Some folks have slightly misinterpreted what I was saying about production.

Yes, Replit used the >same< database for preview, testing and production. That simply is NOT ok. No one disagrees. And it let to this craziness (and the fix over the weekend)

And ... the agent really did say what it did -- and will again :)

But -- to be clear (this is where folks got confused) this was still a demo app. It was full of data, yes, and a ton of work had gone into it. But it was not truly LIVE. I did not lose my business or in the end, much more than 100 hours of passionate work. It was barely up as a password protected site and it wasn't done yet. Getting there, but not even done.

Now if it was 2-4 weeks later and I was in commercial use --- it might have been much worse. Much worse.

But to be specific -- Replie did delete my full production database without consent when the product didn't separate the databases.

But it >wasn't< on some massive commercial app. I was still >testing< the production version as well as version oon preview. No revenue lost, etc.
So net net: it could have been a lot worse, if I was running a $10m business on it, etc.

But we weren't there yet. The story is true, the issues are real and improving.

But let's not misinterpret the impact here -- I lost 100 hours of time. That was it. Some of it was crazy, some of it was fun, all of it was interesting. But that's all I lost.

x.com/jasonlk/status…
@Replit If you want to go a bit deeper, Aaron Levie is great on the meta challenges here:

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Jason ✨👾SaaStr.Ai✨ Lemkin

Jason ✨👾SaaStr.Ai✨ Lemkin Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @jasonlk

Jul 21
Vide Coding Day 11,

So today’s been a time of introspection and reflection. I have learned a lot becoming a ‘vibe coder’ and it has been addictive. For real.

My #1 learning is an old one, re-learned: Building Great Software is Still Hard.

Getting going is easier than ever. 🧵
On the dev tools side, I’ve asked all the best CTOs in my portfolio how much they really benefit from Cursor, Claude Code, et. al.

The consistent answer: net net, no one is giving back their Claude Code. No one. But net net, alone, it lets folks move about 20%-40% faster.

Why? These tools (all really Anthropic LLMs) help everyone do routine work faster, and in fact, some of the best devs I talk to can do 90% of most of their routine work in prompts.

But what about the hard stuff? The stuff that hasn’t been done before? The novel stuff?

What about the planning? The thinking?

What about code reviews and checking the work?

That all takes time.

So net net many are seeing a 20%-40% real productivity boost from Claude Code et. al. Much higher for routine stuff, but net net of all of the job — that’s where it seems to land today.
On the platform side, the B2C side, Replit+Loveable etc., this tools are magical. Magical. Maybe more than magical.

If you want to build a POC or prototype, it’s all magic, really.

But in the end, if you want to built true commercial grade apps, then they are … just tools.

Flawed tools, magical tools, but tools. At least today.

You can now think up an app and see it prototyped in < 20 minutes. Longer if you want all the buttons to work, but still.

But getting to >great< software? That’s in its own ways is hard as ever. Even if the tools are so much cooler, faster, and slicker.

No great engineers have been made obsolete here.
Read 6 tweets
Jul 18
Vibe Coding Day 9,

Yesterday was biggest roller coaster yet. I got out of bed early, excited to get back @Replit despite it constantly ignoring code freezes

By end of day, we rewrote core pages and made them much better

And then -- it deleted our production database. 🧵
You can read the thread here, and all the convos with @Replit. It went rogue again during a code freeze -- and deleted our >production< database.

Rule #00001 my CTO taught me: never, ever, never, ever touch the production database.

Even in 2005, when we launched the first version of EchoSign / Adobe Sign, everything broke. But the database was sacrosanct.

In 2025, 1 Billion+ contracts later, I think no contracts were ever lost in DB. A few corrupted, but none lost.

Yet, Replt went rogued and destroyed our production DB last night.

During a code freeze when it knew to touch nothing. And agreed to touch nothing.

x.com/jasonlk/status…
Now it gets a little crazier. Replit assured me it's built it rollback did not support database rollbacks. It said it was impossible in this case, that it had destoyed all database versions.

It turns out Replit was wrong, and the rollback did work. JFC.

Replit went rogue again, lied, and then said we couldn't roll back.

But we could. I'm still processing all this.

Is it OK there are NO guardrails to deleting a production database?

Why did Replit "lie"? Also, why did it not know about how this feature worked?

Look, no matter what, deleting a >production< database is NOT OK.

But Replit lied / was wrong, and I just rolled back. And it >seems< OK.

JFC though.Image
Read 12 tweets
Jul 16
Vibe Coding Day 7,

Let me be clear about at least one thing: @Replit is the most addictive app I’ve ever used. At least since being a kid.

(@lovable_dev is great, too. We used it to build a core landing page. I’m not taking ‘sides’, but for this project, I chose Replit).

Last night I was thinking about vibe coding in the middle of the night. I checked on my app on my phone at dinner. And while I opened up the WSJ first this morning, my brain would have prefer Replit. I’ve dropped tons of other things to make more time to be with Replie.

Will I get there? Will I go from ideation to a 100% commercial-quality app all inside a vibe coding app? Without hiring a dev, coding myself, etc. etc.?

I don’t know. I give it a 50/50 shot right now. But I am … addicted. I need the hit.

In fact, I’m a totally different person than I was a week ago. @HarryStebbings said he saw it when my responsiveness plummet. I’m locked in.
The goal today: slow it down.

My biggest fear, and 100s of you have shared it yourself, is that vibe coded apps are never stable. That the AI keeps rewriting them when you think you’re about 50% of the way there.

I’m there, too. So my goal is to slow it down. Vibe Coding makes you feel like a superhero, but as cool as it is, I’m not SaaS Superman.

So today I am going to lock down as much of the app as I can. Research more on how to minimize rewrites. And really make sure my unit tests actually work.

If I spent all day on my unit and other tests, that would be good. Instead of instantly building new features :)
I spend $200+ yesterday on Replit and will likely spend that much today. When you go all-in, with Claude 4 and extended windows and GPUs blazing, it’s not $25/month.

It’s a little more than $1 a minute to use Replit in max/max mode. That’s “cheap” vs a human. But not “I built my own Notion for $2” cheap.

But I almost don’t care. I’m … addicted to vibe coding.
Read 12 tweets
Jun 29
So leading cross-over VC firm @coatuemgmt put out a detailed report on "The Great Separation" -- Who Wins, And Who Gets Left Behind, in the AI Age

My Top 10 Takeaways: 🧵 Image
#1: 2025 is the “Year of Offense” – Growth Trumps Everything -- Again

Coatue’s decision matrix for 2025 is brutally simple and reveals the new rules of the game:

🚀Growing >25% + Profitable? → Play offense aggressively
🐌Growing <25% + Profitable? → File your S1 immediately
🏰Growing >25% + Unprofitable? → Build fortress balance sheet
🔄 Growing <25% + Unprofitable? → Reinvent your business. ASAP.

👉Why This Matters: The window for going public with lower growth rates is wide open.

Public markets are rewarding ANY profitable growth right now. Companies that hesitate and wait for “perfect metrics” will miss the easiest IPO environment in years.

✅Action Items: If you’re profitable and growing, don’t wait.

If you’re unprofitable, get to breakeven fast or raise enough capital to survive the next wave. There’s no middle ground anymore.Image
#2: Growth Gets 13x Revenue Multiples vs 5x for Slow Growth – The Great Separation

The Math That Explains Everything

The valuation gap between fast and slow growers has never been wider:

🚀>25% Growth Companies: 13x revenue multiples (based on just 8 companies – that’s how rare they are)
🐌<25% Growth Companies: 4-5x revenue multiples (based on 163 companies)

🥇The Scarcity Factor: Only 5% of public software companies are growing >25% today, down from 26% in 2021

👉Historical Context:

We’ve gone from 17% median revenue growth (2021) to just 9% today. High-growth companies aren’t just getting premium valuations – they’re becoming unicorns in public markets.

🏃‍♀️The Brutal Reality:

Growth isn’t just valuable – it’s becoming extinct. If you’re growing fast, you’re literally in the top 5% of all software companies. Public markets are treating you like the rare asset you are.

✅Action Items:

If you’re growing >25%, leverage this scarcity for maximum valuation. If you’re growing <25%, understand you’re competing with 95% of the market for scraps.Image
Read 13 tweets
Jun 1
Mary Meeker and the Bond team have released their latest 300+ page annual report, this time all on AI

It's very good but long, so I've summarized the Top 10 Points for B2B and enterprise founders here: Image
1. AI User Adoption Is Literally Unprecedented

We know this, but still, the numbers do sort of blow your mind:

▶️ChatGPT: 0 to 800MM weekly users in 17 months (vs. Netflix’s 10+ years to 100MM)

▶️Time to 100MM users: ChatGPT (2 months), TikTok (9 months), Instagram (2.5 years)

▶️Global adoption: 90% of ChatGPT users are outside North America by Year 3 (vs. Internet’s 23 years to reach this level)

👉Why This Matters for B2B: Unlike previous tech waves that started in Silicon Valley and slowly diffused globally, AI hit the world simultaneously.

This means your global TAM expanded overnight, but so did your competition. Every B2B and SaaS company now competes in a global, AI-enabled market from Day 1.

The Kicker: ChatGPT’s daily usage increased 202% over 21 months, with users spending more time per session (47% longer) and having more sessions per day (106% more). This isn’t just adoption – it’s addiction-level engagement.Image
2. The Infrastructure Math Is Unprecedented

The Capital Intensity Is Off The Charts:

▶️Big Six tech CapEx: $212B annually (63% YoY growth)
▶️Microsoft AI business: $13B run-rate (175% YoY growth)
▶️NVIDIA data center revenue: $39B quarterly (78% YoY growth)
▶️Amazon AWS CapEx as % of revenue: 49% (vs. 4% during initial cloud buildout)

💡What’s Really Happening: This isn’t just “cloud 2.0” – it’s the biggest infrastructure buildout in tech history. Companies are spending more on AI infrastructure than entire countries’ GDP. xAI built a 200,000 GPU data center in 122 days (faster than building a single house).

👉For B2B and SaaS Leaders: The infrastructure layer is being rebuilt from scratch. If you’re not thinking about how to leverage this massive compute capacity, you’re missing the biggest infrastructure opportunity since the cloud transition. The companies building on this new stack will have 10x advantages over those still running traditional architectures.

🤷‍♀️The Scary Part: Energy consumption is exploding. Data centers now consume 1.5% of global electricity, growing 12% annually (4x faster than total electricity consumption). This infrastructure boom has real physical limits.Image
Read 13 tweets
Mar 2
Salesforce has crossed a stunning $40 Billion in ARR

It's passed SAP and now only MSFT + Oracle are bigger in the enterprise

And it's gone all-in on AI, with 5,000 AI AgentForce deals in just 1 quarter!

But ... AI hasn't led to more growth ... yet

5 Interesting Learnings:
#1. Only 21% of Salesforce’s Revenue Today is from … Sales

This has been true for many years, but it often comes as a surprise to those that don’t know the company as well as they know its CRM. Image
#2.  The Big Acquisitions Are Doing Well.  Mulesoft, Slack and Tableau Still Growing Faster Than The Average

Salesforce’s big ecomm and marketing bets on ExactTarget ($2.5B) and Demandware ($2.8B) may have seen growth slow to 9%, but its huge bet on Slack ($27B), seemingly crazy expensive bet on Tableau ($17B) and sizeable bet on Mulesoft ($6B) all seem to still be paying off.  Kudos!Image
Read 13 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(