fanioz

Building in public.
I automate digital workflows, share what actually works, and document the experiments along the way.

[fanioz @faniozilla]

I build with AI tools.

Currently exploring PaperclipAI. https://t.co/alYfbcBIU7

Sharing what actually works vs. what's just hype.

[fanioz @faniozilla]

@aronprins @papercliping @dotta @devinfoley @ScottTong congratualtion...you deserve it

[fanioz @faniozilla]

@shivsakhuja goose

[Paperclip @papercliping]

Paperclip v2026.609.0 dropped

💸 Anthropic Fable 5 support
🖼️ Company Artifacts and Rich Attachments
☑️ Checkbox Confirmations
🛡️ Low-trust agents

https://t.co/AOGbP7IffT

[fanioz @faniozilla]

@cjzafir That's impressive. May I try on M4 24GB?

[fanioz @faniozilla]

Just open-sourced SysMonitor! It's a native macOS system monitor that lives right in your menu bar. Built it because I wanted a beautiful, low-overhead way to keep an eye on my
Mac's performance without the clutter.

Most monitors either give you too little info or overwhelm you with giant dashboards. SysMonitor gives you a quick glance in the menu bar, then drops down a translucent
glassmorphism widget when you click it.

Under the hood, it's 100% Swift and SwiftUI. Getting the window to snap perfectly under the menu bar icon took some math with `NSStatusItem` frames, but the result feels completely native. It even auto-hides when you click away.

It tracks CPU (per-core), RAM breakdown, Disk I/O, and Network speeds. Best part? It automatically throttles its own polling when the widget is hidden so it won't drain your
battery in the background.

The code is up on GitHub if you want to build it yourself, or if you just want to see how the glassmorphism UI is put together.
Repo: https://t.co/7C5qJfNa2P

[fanioz @faniozilla]

impressive !

It's the combination of a capable model Grok Build 0.1 and KiloCode's workflow: plan → code → review.

Give AI a structured process, and suddenly a webhook delivery service gets planned, built, and shipped from a single prompt. (plan mode I guess)

The model matters.

The workflow matters just as much.

[Kilo @kilocode]

We gave Grok Build 0.1 one prompt: build a webhook delivery service in TypeScript, Bun, and SQLite.

It planned it, built it, and shipped a working demo.
Total cost: $1.65.

Zero tool-calling failures.
Here's exactly what happened, file by file.

[fanioz @faniozilla]

@dabit3 @DevinAI Im buiding mobile app to learn English

[fanioz @faniozilla]

another weekend experiment.

genuinely curious how far this thing can go:

looks like people are trying to make Claude-style coding workflows more accessible/open. not sure yet if this is actually useful long term or just another “AI wrapper with vibes”

but the repo is moving fast enough that i kinda want to test it myself.

[fanioz @faniozilla]

vibecoding is fun until you realize half the apps getting shipped are basically prompt spaghetti held together by pure optimism 😭

but honestly… this is probably the first time non-devs can build something actually usable without spending years learning engineering first.

and that changes a lot.

what’s interesting to me isn’t “AI replacing programmers.”
it’s the fact people suddenly feel confident enough to build weird little ideas that used to stay buried in their notes app forever.

and weirdly enough,
the vibecoded products that survive usually aren’t the cleanest technically.

they’re just the ones with the fastest feedback loop.

ship fast.
break stuff.
rewrite later.
somehow it works.

unpopular opinion maybe:
vibecoding is not the end of software engineering.

if anything, it makes good engineering way more obvious.
because once everyone can generate code, structure and decision-making become the real skill.

curious how many people here have actually shipped something with vibecoding already?

[fanioz @faniozilla]

@dotta wow thats colorful

[Daniel Ch @chddaniel]

BREAKING: ai slop is now 100% optional.

Introducing Shipper - the first AI business builder to replace a 9-5 salary.

It’s the foundation for a one-person billion dollar company, coordinating AI workers across coding, sales, marketing, operations, and design.

🚨 Giving away unlimited FREE credits for the next 24 hours, randomly.

(until they run out)

Want access?

• Retweet
• Leave a 'Shipper' comment

[fanioz @faniozilla]

@chddaniel yoo...shipping with shipper

[fanioz @faniozilla]

I want to start experimenting with WUPHF

at first I thought it was just a chaotic notification app inspired by Ryan’s startup from The Office 😭
turns out… not really.
it’s more like an “office” for AI agents.

Claude Code, Codex, OpenClaw, local LLMs, all working together inside one shared workspace with shared memory, so agents don’t have to restart context from zero every time.

the part that got my attention is this:
agents can collaborate and maintain their own knowledge base almost like a small team.

honestly the whole “AI office” concept still feels kinda surreal to me.
but I can already see interesting use cases for:
— automation workflows
— coding tasks
— second brain systems
— autonomous project management
and yeah… the WUPHF name is still hilarious 💀

curious if anyone here is actually using it in real workflows, not just demos.

[fanioz @faniozilla]

idk why but this still feels kinda unreal to me 🧐

“build native Swift iOS apps directly on your phone” sounds insane enough already.

but running it in a real simulator + install directly on device + share to 100+ people?

either Apple quietly changed something big… or there’s a very creative workaround happening behind the scenes

[fanioz @faniozilla]

@anything this is good, I want it

[fanioz @faniozilla]

@kilocode @Steve_Yegge hello, why I cant scroll the output text by Mayor?

[fanioz @faniozilla]

Gas Town is interesting not because “AI agents can talk to each other.”

That part is already becoming normal.

What makes me curious is the idea of giving multiple agents a shared environment, where they can debate, coordinate, form opinions, maybe even develop weird social dynamics on their own.

And honestly… I still can’t tell if this is a gimmick or something genuinely useful 😭

I want to start experimenting with Gas Town
but the more I read the docs, the less sure I am about the actual killer use case.

So far I’ve thought about:

— autonomous research teams
— simulated customer interviews
— AI company sandbox
— testing emergent agent behavior

but most of them still feel very “tech demo” to me. Not “wow this is actually valuable”.

This is something I keep noticing lately:
a lot of AI infra is technically impressive, but the real-world use case still feels… unresolved.

Maybe I’m missing something.

Guys, If you had to build something with Gas Town, what would you make?

[Kilo @kilocode]

The factories can talk to each other now.

Gas Town by Kilo is generally available today, with the Wasteland built in. @Steve_Yegge's multi-agent orchestrator, hosted on Kilo Cloud, federated across thousands of towns.

One Mayor. Hundreds of agents. 500+ models. Zero infrastructure to babysit.

This is what agentic engineering looks like when you stop pretending one developer plus one assistant is the ceiling.

https://t.co/amOMOyhxEZ

[fanioz @faniozilla]

@gregpr07 is that codex that create those .png files?

[fanioz @faniozilla]

@aronprins yes.. love this

[fanioz @faniozilla]

Plot twist.

I just cancelled a product my AI CEO proposed.

Turns out I had already rejected it 3 weeks ago.

I forgot. Approved it anyway. Nobody caught it.

This seems like what "founder fatigue" looks like in an
AI-powered company:

The agents don't remember your previous decisions
across sessions. They propose based on research.

You approve based on... whatever state you're in
that day.

Lesson learned (again):

Before approving anything, check the history.
Your agents can't do that for you.

The human memory is still the weakest link.