Thomas Langhorst

376 posts

Thomas Langhorst banner
Thomas Langhorst

Thomas Langhorst

@the_langhorst

Software Engineer | YouTuber. I tweet about my journey of building a Side Hustle.

Katılım Mart 2019
101 Takip Edilen47 Takipçiler
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@jonathanbrnd As an engineer learning how to market I think I am too biased to answer this question 🤣
English
0
0
1
7
Jonathan
Jonathan@jonathanbrnd·
who wins? 1. a cracked marketer learning vibecoding 2. a cracked engineer learning marketing
English
713
38
2.3K
317.7K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
Big lesson while building my AI-for-devs guide: Setup > prompting. AI isn’t inconsistent — it’s stateless. Every new chat starts from zero. Persistent context changed everything. Building this in public.
English
0
0
0
6
shirish
shirish@shiri_shh·
dear algo please only show this tweet to S tier programmers
English
669
66
3.6K
175.6K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
One simple step reduced AI mistakes for me and my team: We stopped over-specifying. Instead of: "Create a service, add this endpoint, use these functions, follow these steps…" We used: "Add an endpoint that follows our auth pattern and matches existing error handling."
English
1
0
0
12
Thomas Langhorst
Thomas Langhorst@the_langhorst·
You need this if you use AI on a real codebase: One shared instructions file. Little known fact about what to put in this file: - Architecture decisions. - Conventions. - What NOT to do. It's onboarding for humans *and* AI!
English
0
0
0
18
David Ch
David Ch@chhddavid·
day 143 of building in public MRR: $5.3k Goal: $10k still working hard on gathering templates. i've made some myself, with a mix of what other people have built so amazing to see what Shipper can do! goal is to pick the most beautiful and functional apps so that the 1st impression is "wow this tool can actually build cool stuff"
David Ch@chhddavid

made in Shipper

English
10
0
20
2.2K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@starter_story Good advice! Thanks for sharing. I find it hard though. I have a gazillion posts und mind, but none of them work. What do you think works best: sharing mistakes, accomplishments or just simple advise from “today I learned”?
English
0
0
0
13
Starter Story
Starter Story@starter_story·
tl;dr You should do cool, tangible things, and share them. Most people on X post recycled advice or ChatGPT slop. None of that signals progress or competence. It’s abstract and forgettable. There’s no reason to follow someone who isn’t actually doing anything. AUDIENCE GROWTH COMES FROM VISIBLE OUTPUT Additioanlly: progress, mistakes, and experiments feel real, and people connect with that. It also lowers the fear of posting. The reality: no one is really paying much attention to you; they’re focused on themselves! Thanks @jackfriks for the awesome interview!
English
6
1
41
4.4K
Starter Story
Starter Story@starter_story·
$20K MRR founder shares what most people get wrong about "building in public":
English
43
26
550
35.1K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@MakadiaHarsh So true! Getting up at 4am because the server crashed … this is what most people don’t have in mind when they hear “ship fast, iterate”
English
1
0
1
18
Harsh Makadia
Harsh Makadia@MakadiaHarsh·
The "ship fast, iterate" playbook works, but most founders skip the unglamorous middle. The daily user calls, the constant feedback loops, the obsessive focus on actual problems before they even think about scaling. You can ship an MVP fast. You can get interest, a waitlist, even early revenue. If your product is simple or you have basic product instincts, sometimes you really can build and validate something meaningful quickly. It happens. But most of the time real products still take months. They need iteration, debugging, new features, and actual users so you can test and fix. Non-technical founders often miss that products are living organisms. They need maintenance, they need oxygen (distribution and retention), and none of that is instant or one-time work. Anyone claiming you can automate user research is lying. Anyone promising product-market fit in 2 weeks is selling an edge case. These outcomes happen, but they're rare and treating them as the standard is how founders end up burned out and bitter. A lot of the "just ship it" advice comes from comparing today to the 2000s/2010s, when companies were built with decades in mind. Too many founders now fantasize about quick wins instead of deep user understanding, largely due to X and very largely due to every guru jamming the "launch in a weekend" narrative down our throats. I'm not saying you shouldn't want to move fast. That's actually exactly what i prefer. You might see more success by aiming for user connection first because the opposite is a very unlikely-to-hit gamble where you only see the wins on your feed. Do not let X convince you that shipping fast means skipping the conversations or that 10k users in a month without talking to any of them is normal.
English
6
0
13
1.4K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@casinokrisa @alexcooldev So true! I am able to fix bugs in minutes now, but I know exactly where to look and which routes to neglect. Knowing how to code gives me the unfair advantage
English
1
0
0
268
Mikhail Drozdov
Mikhail Drozdov@casinokrisa·
@alexcooldev Vibe-coding works until it hits scaling or debugging walls that require solid fundamentals.
English
10
1
41
4.9K
Alex Nguyen
Alex Nguyen@alexcooldev·
To be honest, I'm also vibe-coding 80–90% of all my projects. Vibe-coding feels amazing when you actually know how to code and understand how the system works.
English
356
231
5K
195.1K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@alexcooldev Absolutely! I feel like I have a junior dev doing all the work while I get to do the cool parts 💪
English
0
0
0
6
Thomas Langhorst
Thomas Langhorst@the_langhorst·
I learned this the hard way: I always expected Copilot to *remember* things. It doesn't. If it’s important, it belongs in an instructions file — not in last week's chat. Let Copilot update your instructions files regularely.
English
0
0
0
13
Thomas Langhorst
Thomas Langhorst@the_langhorst·
Small lesson I learned the hard way: Generic AI instructions make output worse, not better. Specific rules > more context. Once I stopped saying "follow best practices" and started saying "use Zod for type validation, see file <path_to_file>", AI finally became useful.
English
0
0
0
6
Thomas Langhorst
Thomas Langhorst@the_langhorst·
Most teams misuse AI because they confuse context with instructions. Context = what AI knows Instructions = how it should behave Once I separated the two, hallucinations dropped fast.
English
0
0
0
8
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@alexcooldev So true! This is why I have a daily power hour. I build or create something for at least 1 hour a day. This keeps me rolling, no matter the mood or motivation.
English
0
0
0
12
Alex Nguyen
Alex Nguyen@alexcooldev·
Stop overthinking. Build something. A site. An API. A messy MVP. A Chrome extension. An AI agent. A tiny automation. Momentum changes everything.
English
308
500
5.6K
134.6K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@corbin_braun Totally true! I build a while MVP application in under 2 hours. Before AI, this would have taken me at least a whole weekend
English
0
0
0
30
corbin
corbin@corbin_braun·
you can build anything now the only gap is time and knowledge AI is your workforce
English
60
6
146
4.1K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
Most engineering leaders worry about AI adoption. Here's a mindset shift that helped me: Don't tell devs to 'use AI more'. Instead, suggest moments in their workflow where AI creates instant relief. Relief → trust → adoption. It's never the other way around.
English
0
0
0
7
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@cleanwithmike I am still working a 9-5, but I still started doing this. Now I get to spend time with my wife, enjoy nice walks and play video games, while AI + automations (+ a few own written software pieces) do a lot of work in the background.
English
0
0
0
8
Mike Cleans
Mike Cleans@cleanwithmike·
Step 5: Bought back my time. This is the step most people skip. Making money is great. But if you're working 60 hours a week, you've just traded one prison for another. I spent 9 months finding the right virtual assistants and building systems.
English
3
0
14
4K
Mike Cleans
Mike Cleans@cleanwithmike·
I’m 32 & make $40k/month… But 4 years ago, I was: • Lost • Broke • Anxious Here are the 5 steps I took to reinvent myself (so you can too):
Mike Cleans tweet media
English
33
46
742
112.7K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@hayesdev_ Thanks for sharing! I now know what to watch for my coffee break 😅
English
0
0
0
635
Julian Goldie SEO
Julian Goldie SEO@JulianGoldieSEO·
This update quietly changed coding forever. Here’s the new flow ⤵️   1. Install Chrome DevTools MCP   2. Connect Cursor or Claude or Copilot   3. AI launches Chrome   4. AI inspects DOM + console   5. AI fixes and validates bugs No screenshots. No guessing. Save this video, you’ll automate debugging. Want the SOP? DM me. 💬
English
16
38
372
28.9K
Thomas Langhorst
Thomas Langhorst@the_langhorst·
@andi_losing Optimizing the wrong thing is always a lingering danger. Glad to hear you overcame this 💪
English
0
0
1
8
Andi
Andi@andi_losing·
I made a lot of mistakes in the past weeks I wasted days setting up my own VPS I wasted days switching to Supabase I learned a lesson I should optimize my shipping speed, not my tech stack
English
51
2
97
3.1K