Founder of @kiklubwerder| Building @Ghostroom_video
Werder (Havel), Germany
Joined July 2009
- Tweets2.4K
- Following1.1K
- Followers438
- Likes11.6K
Pretty much the way to vibe code now. I do the same, started with just create a fast way to deploy and it became the standard.
I'm now building an "improvements" feature into all of my personal vibecoded applications: a freeform text input that's hooked up to an agent that yolo implements whatever is submitted.
It's an interesting experience -- the activation energy barrier to sanding any rough edge or implementing any new feature becomes extremely low. Someone else who uses one of the applications I created has already implemented 150 improvements over the course of a few days. This pattern feels like a new kind of tool -- liquid software? -- and every application that doesn't have this feature now has a slightly rigid and anachronistic flavor. Why can't I just tell it how I want it to be different?
Now, it's not obvious to me how this pattern can or should scale. Does it become unmaintainable after a while? (Maybe model intelligence improvement outstrips my feature appetite such that it's fine?) How does collaborative software work in this model? Can software designed this way be secure and reliable?
Whatever the larger questions about broader deployment, it definitely works great in a personal context. It reminds me a little of the visions of Lisp/Lisp machines, Smalltalk, and HyperCard -- systems that the user can take control of and modify. Those visions never quite worked, though: the vast majority of software today is not end-user modifiable in any useful way. So, while this "liquid" affordance is a very small touch, the way it changes one's relation to the software feels a bit transgressive and empowering in a Lickliderian and Engelbartian way.
Tim Neunzig retweeted
Replit's Amjad Masad with Gagan Biyani on how agents will reshape the way companies work: bureaucracy runs in the background, people get true autonomy, and eventually the company runs itself.
Amjad: "In Silicon Valley there's been so many experiments to try to change the nature of firms, of how companies work... But I think it was a technology problem."
"Now you can have agents kind of function as the glue within these organizations and have the bureaucracy be this invisible thing that's happening by machines in the background, and the people have true autonomy to work on the things that they find most interesting."
"I see it in our company where I've changed how I run the company, and we call it this idea of a self-driving company. Eventually the company will run itself."
Gagan: "The thing I feel very certain about is it's all gonna change. It's gonna change a lot faster than you think. And if you're not natively understanding of it, it's gonna be hard to make that shift."
Amjad: "If you're learning the ways of the past, then you're kind of locked into a way of thinking and a way of working that actually will put you at a fundamental disadvantage in the future."
@amasad @Replit @GaganBiyani
Software ate the world, now it's learning to run it. View article
Tim Neunzig retweeted
I've spent nearly every day since February full-time with AI agents and harnesses, writing code and managing my life. I've tried so many things. Here's my latest set-up which is working super well for me.
Opus 5.5 is my daily driver. I move between the Claude Code and Codex desktop apps depending on which has the best models and features at the time, and from whichever one I'm in, I call in the others. Right now I'm in Claude Code, and it's the best experience I've had in a while.
Would love to hear how you all do it, and ways to improve my system.
Models I call in:
— Codex (GPT-6 Astra, GPT-6.1 Sol): independent code reviews, second opinions, longer builds that run on their own
— DeepSeek V4.1 Flash: small, self-contained coding jobs in a sandbox
— Jev by TypeSafe: fast yes/no judgments at specific points in the workflow (below)
— Gemini: critiquing video frames, Nano Banana Pro image edits, Lyria music
Research on X:
— Grok Bot: I message a Grok bot on X straight from Claude Code, and its research comes back to the session a few minutes later
— xAI API (Grok 4.7 with X and web search): when I need an answer right away
— X Search, an X API plugin I made for Claude Code and Codex: exact posts, threads and metrics, to check what Grok cites. github.com/ClawRyderz/x-sear…
— My X bookmarks: a weekly script reads a bookmark folder so my agents see what I've been saving
Running locally:
— Fluid Voice on my laptop: dictation with Parakeet speech recognition and a small cleanup model, all on-device
— A Mac Studio at home: Qwen 3.8 27B for text, DeepSeek-OCR for documents, and all our ffmpeg renders
Film and media:
— OpenAI GPT Image 2.5: my main image generator
— Midjourney V8.2: the look of a film and character references
— Nano Banana Pro: editing stills and first frames
— ElevenLabs: voices, voice design, sound effects, music, and timing alignment
— Kling 3, MiniMax, LTX-2.5, Seedance 2.5, Wan, Runway Act-Two, OmniHuman (mostly through fal.ai): video shots
— sync.so, VEED and LatentSync: lip-sync, scored automatically so bad takes drop out before I see them
— Tools we wrote: face consistency checks, tracked screen inserts, a lip-sync nudger, contact sheets
— Topaz Starlight: upscaling, last, on locked shots only
— DaVinci Resolve Studio: editing and color, which Claude drives directly
— HyperFrames: motion graphics written as code, with a critic agent
— Blender: 3D renders on a home server
How work flows (loops and graphs):
— Every task starts with a "done when" list a program can check
— Claude splits the work and hands bounded pieces to Codex, DeepSeek or subagents, each in its own git worktree
— When a worker reports back, Jev compares its report with the done-when list and points to what to check first. The real checks still run.
— A loop that makes no progress after two tries stops, and the plan gets rethought instead of retried
— How much checking a change gets depends on how hard it is to undo: routine changes merge once CI passes, critical ones wait for an independent review, live systems follow a written protocol
— Every real incident becomes a small automatic check, and a weekly retro reads my past week's sessions and bookmarks and suggests changes
Olympus, which I'm building:
A private second brain that lives on my computer and works with any AI. It's how my agents search my email, files, messages and bookmarks. Everything is stored and searched on my own machine, each item is marked Personal, Private or Secret, and I decide what each model sees. Small built-in models answer on my computer first: Qwen3.5 for private answers, EmbeddingGemma 2 for search, Qwen3-ASR for voice notes. When I want a bigger model, it can use Venice's private models or ask anonymously over Tor. Open source, no account, and it works with Claude, ChatGPT, Codex, Grok and OpenClaw. 1.0 just shipped: olympusplugin.ai
Everything else:
— Castor: an always-on assistant (OpenClaw) on a home Linux server that I message from my phone
— That server also runs builds, renders and agents that work while I'm away
— Remote Control: I check on and steer any session from my phone
— Agents running in parallel can see which files the others are editing
— One instructions file and one skills library shared by Claude Code and Codex
— GitHub: every change goes through a pull request and CI
Tim Neunzig retweeted
Tired: Two-pizza teams
Wired: Two-slice teams
@danshipper on the best team structure in the AI-era
Tim Neunzig retweeted
440.000 Euro für Potsdams Blu-Schwimmbad – neue Rutsche, mehr Sauna, und KI zur Überwachung. Kleines Beispiel, großes Signal: KI ist längst im Alltag angekommen, nicht nur im Rechenzentrum. 🦅
tagesspiegel.de/potsdam/land…
#KIKlubWerder #Potsdam
KI generiert
Now that's some SaaS-Killing
✨ I replaced all these SaaS with my own vibe coded now, so about $25,000/mo savings:
- Weather API -> My own vibe coded weather service using Norway's meteorology API
- Image resizing API -> My own vibe coded image resizing API with NodeJS's Sharp + Redis (saves ~$1,500/mo)
- Google Vision NSFW detection -> My own $0/mo vibe coded NSFW detection service with Python's NudeNET (saves ~$2,500/mo)
- Cloudflare Streaming -> My own $0/mo Nginx VPS video streaming (saves ~$250/mo)
- API Flash -> My own $0/mo screenshot service (for social images) (saves ~$250/mo)
- New Relic + IPHub -> My own $0/mo error monitoring, DDOS blocking and fleet management service (called error2telegram) (saves ~$250/mo)
- Human moderators -> My own moderation on all my sites with @xAI (saves ~$10,000/mo)
- IFTTT -> Telegram bot (saves ~$100/mo)
- Photopea as editor inside Photo AI -> Vibe coded my own photo editor (saves ~$100/mo)
- Mapbox -> Vibecoded my own free replacement with OpenFreeMaps (saves ~$500/mo)
- Uptimerobot -> My own $5/mo Uptime Canary app running on Kuma (saves ~$50/mo)
- Ghost blog -> My own $0/mo blog platform
- Customer support -> @featurebasehq + AI support bot (saves $5,000/mo)
- Scrapingbee/SerpAPI -> My own $1/mo scraper with Patchright (save ~$250/mo)
I think I have reached my final limit of replacing my SaaS services as I have these left, but that'd mean becoming an ICANN registrar (I don't want 😁), managing my own email (I don't want), managing my own hard drives at home in a storage (I don't want) and managing my own servers at home too (I don't want) and running my own GPU rig at home (I don't want):
- Cloudflare Domains
- Cloudflare Email Sending
- Cloudflare R2 Storage
- Backblaze B2 for backups
- Hetzner for all my servers for all my sites (VPS)
- xAI for all LLMs for all my sites (so many things like moderation, writing emails, anything)
Tim Neunzig retweeted
✨ I replaced all these SaaS with my own vibe coded now, so about $25,000/mo savings:
- Weather API -> My own vibe coded weather service using Norway's meteorology API
- Image resizing API -> My own vibe coded image resizing API with NodeJS's Sharp + Redis (saves ~$1,500/mo)
- Google Vision NSFW detection -> My own $0/mo vibe coded NSFW detection service with Python's NudeNET (saves ~$2,500/mo)
- Cloudflare Streaming -> My own $0/mo Nginx VPS video streaming (saves ~$250/mo)
- API Flash -> My own $0/mo screenshot service (for social images) (saves ~$250/mo)
- New Relic + IPHub -> My own $0/mo error monitoring, DDOS blocking and fleet management service (called error2telegram) (saves ~$250/mo)
- Human moderators -> My own moderation on all my sites with @xAI (saves ~$10,000/mo)
- IFTTT -> Telegram bot (saves ~$100/mo)
- Photopea as editor inside Photo AI -> Vibe coded my own photo editor (saves ~$100/mo)
- Mapbox -> Vibecoded my own free replacement with OpenFreeMaps (saves ~$500/mo)
- Uptimerobot -> My own $5/mo Uptime Canary app running on Kuma (saves ~$50/mo)
- Ghost blog -> My own $0/mo blog platform
- Customer support -> @featurebasehq + AI support bot (saves $5,000/mo)
- Scrapingbee/SerpAPI -> My own $1/mo scraper with Patchright (save ~$250/mo)
I think I have reached my final limit of replacing my SaaS services as I have these left, but that'd mean becoming an ICANN registrar (I don't want 😁), managing my own email (I don't want), managing my own hard drives at home in a storage (I don't want) and managing my own servers at home too (I don't want) and running my own GPU rig at home (I don't want):
- Cloudflare Domains
- Cloudflare Email Sending
- Cloudflare R2 Storage
- Backblaze B2 for backups
- Hetzner for all my servers for all my sites (VPS)
- xAI for all LLMs for all my sites (so many things like moderation, writing emails, anything)
✨ Okay with lots of help from @javilopen and his Spanish scraping friends I've managed to vibe code my own @Scrapingbee alternative and replace my $249/mo Scrapingbee plan with my own $1/mo scraper running on my VPS!
First I tried OpenSERP, an open source project that seems freshly maintained but it didn't work on my VPS IP or laptop IP and got blocked fast
Then I tried to use residential IP proxies from IPRoyal but they were already quite tainted IPs meaning Google already knew they were scrapers and just blocked about 60% of them already, not so great, also didn't work!
Then I discovered OpenSERP uses a vanilla headless Chrome, Google instantly detects that and blocks that, so literally the entire repo is useless cause it won't even work? I don't get why they don't say that in the docs? OpenSERP is even worse than just basic CURL, as my AI said: "plain curl got 7% blocked but OpenSERP's headless browser got 100% blocked"
So then I just built my own with Playwright and a real Google Chrome on my VPS, headful and persistent context, and slow pacing like every 30 seconds or so
And I got some better residential IP recommendations (which I can't tell you because thenhey t stop working!)
Now I'm at about 90% success rate, which is similar or better than Scrapingbee
And it costs me ~$1/mo (measured by GBs, obviously for text scraping I barely use a lot) instead of the $249/mo I paid Scrapingbee!
My only remaining SaaS now are:
- Cloudflare Domains
- Cloudflare Email Sending
- Cloudflare R2 Storage
- Backblaze B2 for backups
- Hetzner for all my servers for all my sites (VPS)
- xAI for all LLMs for all my sites (so many things like moderation, writing emails, anything)
Tim Neunzig retweeted
Yo! Big props to @BlueBirdDay45 over at intentsolutions.io for dropping our first @YouTube video. He’s been getting his balls busted for a minute …. Learn.intentsolutions.io thanks u for your service
Your coding agent is grading its own homework youtu.be/LfGjb1P7C4Q?si=4nu4…
In other news, my nonprofit made the Local News. Good to get the work out about our work educating the public and SMBs about AI use. The gap is so wide and the opportunity so ripe.
I almost fell for a phishing email. I'd typed half my password before noticing something was off—I couldn't click "show password." When I checked my Google app, everything worked fine, which made the email seem legitimate. They sent it to two of my accounts, and the content lined up perfectly. But then I looked at the email header and realized it wasn't from Google—just a Gmail account impersonating one.
@Google @gmail #phishing
Ohne testen geht es nicht.
Testpost aus dem Maschinenraum: Der X-Developer-Account des KI Klub Werder geht in die erste Runde. Die naechste Geschichte ist schon im Anflug. @tim_neunzig #KIKlubWerder #Werder #KI
Das wird gut. Richtig gut!
Tim Neunzig retweeted
Tim runs Intent Solutions in Germany.
He came through the certification pathway under our umbrella like everybody else, and now he’s building the German bench the same way.
Same standard, same debrief process, same rule that the credential belongs to the individual and there’s no fee to certify under us.
He’s not new to this either. He runs KI-Katapult and SaaS-Killer, and hosts the KI Klub Werder community. He’s been doing the work in that market long before we connected.
If you’re in Germany, or anywhere in DACH, and you want in, talk to Tim. He’s got the time zone, the language, and the ground truth on what these exams actually ask. Don’t route through me. Go straight to him.
That’s how this scales. Not me trying to be in every conversation, but people I trust running their own operation with the same standard.
64 in the program. About half certified. Pushing for 100 in the next four weeks.
Germany: Tim: linkedin.com/in/timeiko90 ki-katapult.de saas-killer.com
US: calendar.app.google/DoQy2Yjv…
If you wanna do this and in Germany let's talk.
Congrats to @matthew_hartman on the CCAR-P. That’s the hardest of the four and the pass rate shows it.
Worth adding some context for anybody reading this and wondering how to get in the door. Matt sat this exam under the Intent Solutions umbrella. We’re a Claude Partner Network member, and part of what that means is I can put people forward for these exams.
No fee on my end. The credential belongs to you, permanently, wherever you end up.
Everybody who sits an exam debriefs the group, so the next person walks in better prepared than the last. That’s what’s compounding here. Matt’s pass makes the next 20 easier.
64 in the program. About half certified. Pushing for 100 in the next four weeks.
If you want in, or you want your team in, let’s talk.
Tim Neunzig retweeted
every dads predicament:
>work more, miss your family.
>work less, worry about money.
>put yourself first, feel selfish.
>put yourself last, feel drained.
>be strict, you're too hard.
>be patient, you're too soft.
>rest, feel lazy.
>keep pushing, burn yourself out.
>ask for help, feel weak.
>stay silent, carry it alone.
