DEV Community: JackAltman

Just surpassed 5k users

JackAltman — Wed, 01 Apr 2026 13:49:42 +0000

JackAltman

Mar 4

I launched a tiny macOS tool yesterday and woke up to 100+ installs

#news #ai #llm #development

Comments 4

1 min read

The MacBook setup I use as a “vibe coder” (tools that actually matter)

JackAltman — Tue, 31 Mar 2026 23:50:44 +0000

Most “top tools” lists are noise.

If you’re actually building things—shipping fast, testing ideas, running AI workflows—you don’t need 50 apps. You need a tight setup that removes friction.

This is what I’ve ended up using daily.

1. Raycast (replace Spotlight completely)

If you’re still using Spotlight, you’re wasting time.

Raycast becomes your control center:

Open apps instantly
Run scripts
Manage clipboard history
Trigger workflows

It’s one of those tools where after a week, you can’t go back.

2. Arc Browser (context > tabs)

Arc changes how you think about browsing:

Spaces for different contexts (work, side projects, etc.)
Clean UI, no tab chaos
Feels closer to an IDE than a browser

If you’re juggling multiple projects, this matters more than you think.

3. TokenBar (this one fixed a real problem)

If you’re using AI tools a lot (OpenAI, Anthropic, etc.), costs can get out of control fast.

That’s the problem I ran into.

TokenBar:

Lives in your Mac menu bar
Shows real-time token usage + cost
Tracks usage per session/day
Helps you catch runaway prompts before they burn money

It’s simple, but it solves something most people don’t notice until it’s too late.

If you’re building with AI APIs, you probably need this:
👉 https://tokenbar.site

4. iTerm + Oh My Zsh (don’t ignore your terminal)

Default terminal is fine, but upgrading gives you:

Better autocomplete
Cleaner prompts
Faster navigation

Small improvements compound when you’re in terminal all day.

5. Notion (or any lightweight brain dump system)

Doesn’t matter what you use.

What matters:

Quick capture of ideas
Lightweight planning
No friction

Over-structuring kills momentum.

The actual takeaway

The goal isn’t “more tools.”

It’s:

Less friction
Faster iteration
Awareness (especially with AI costs)

Most people optimize for aesthetics.
Good builders optimize for feedback loops.

If you’ve got other tools that genuinely improved your workflow (not just trendy ones), I’m interested.

I built a Mac app that blocks the exact parts of apps that waste your time

JackAltman — Mon, 30 Mar 2026 20:14:33 +0000

Most “productivity” tools don’t work.

Not because they’re bad — but because they’re too extreme.

They either:

block everything (so you quit using them), or
do nothing meaningful

The real problem is smaller.

It’s not YouTube.
It’s the homepage.
It’s not Twitter.
It’s the For You feed.

That’s where time disappears.

So I built something different

Instead of blocking entire apps, this tool lets you block just the addictive parts.

Hide YouTube home + recommendations
Remove infinite feeds (Twitter, etc.)
Keep search, subscriptions, and actual intent-based use

You can still use the internet — just without the traps.

Why this actually works

Because it doesn’t rely on willpower.

It changes the environment.

No dopamine loops → no scrolling → no wasted hours

Other things it does
Schedule “monk mode” sessions
Integrates with Screen Time / system controls
Lightweight (runs quietly in the background)
Who this is for
Developers who open YouTube “for one thing”
Anyone trying to focus but keeps getting pulled into feeds
People who’ve tried blockers and hated them
Free for a bit

I’m letting people try it and giving early users a discount.

👉 https://mac.monk-mode.lifestyle
Use code: DEV

If you try it, I’m curious:
what’s the one site that wastes most of your time?

TokenBar — a simple way to track tokens, costs, and usage across AI models

JackAltman — Tue, 24 Mar 2026 16:13:36 +0000

Working with AI APIs gets expensive quickly.

Between prompts, completions, embeddings, and multiple providers, it becomes difficult to answer basic questions:

How many tokens am I using?
Which feature is driving costs?
Which model is the most efficient?

Most platforms don’t make this easy to track in one place.

What is TokenBar?

TokenBar is a tool for monitoring token usage and cost across AI applications.

It provides a clearer view of:

token consumption
API costs
model usage patterns
Why this matters

As soon as an app starts using models like:

GPT-style APIs
embeddings
multi-model pipelines

cost visibility becomes important.

Without tracking, it’s easy to:

overspend
miss inefficiencies
scale blindly
What it focuses on

Usage visibility
Understand where tokens are being used.
Cost awareness
See how usage translates into actual spend.
Simplicity
Avoid complex dashboards and unnecessary features.

Who it’s for
Developers building AI products
Indie hackers experimenting with APIs
Startups managing inference costs
Anyone working with token-based pricing
Try it

https://tokenbar.site

Summary

AI costs scale with usage.

Tracking tokens and spend early helps avoid unnecessary overhead and improves efficiency.

MetricSync for GERD — track trigger foods without manual logging

JackAltman — Sun, 22 Mar 2026 19:29:00 +0000

Managing GERD often comes down to one thing: figuring out what triggers symptoms.

The problem is consistency.

Meals are hard to track
Portions are easy to underestimate
Patterns are difficult to notice over time

Most people stop logging because it takes too much effort.

What is MetricSync for GERD?

MetricSync is an AI-powered food tracking tool that helps identify potential trigger foods by removing manual logging.

Instead of typing meals, you:

take a photo
let AI identify the food
get instant nutrition estimates
Why this matters for GERD

GERD symptoms are highly individual.

Common triggers include:

high-fat foods
acidic foods
spicy meals
large portions

But identifying personal triggers requires consistent tracking over time.

MetricSync lowers the effort required to do that.

How it works
Take a picture of your meal
AI analyzes the food
Get nutrition data instantly

Over time, this helps you:

recognize patterns
connect meals with symptoms
make more informed food choices
Key benefit

The main advantage is reduced friction.

No:

manual food searches
repetitive logging
complex tracking systems
Important note

This is a support tool, not a medical diagnostic system.

It should not replace:

professional medical advice
prescribed treatment plans
Pricing
Free for 3 days
Then $5/month
Try it

https://metricsync.download/gerd

Summary

For GERD, identifying triggers is critical.

Making food tracking easier increases the likelihood of staying consistent.

MetricSync for Diabetes — using AI to make food tracking actually usable

JackAltman — Sun, 22 Mar 2026 19:28:23 +0000

Managing diabetes isn’t just about medication — it’s about daily decisions.

And one of the hardest parts is food.

What did I just eat?
How many carbs was that?
Is this going to spike my blood sugar?

Most tools expect manual logging, which is slow and often inaccurate.

What is MetricSync for Diabetes?

MetricSync is an AI-powered food tracking tool designed to make nutrition tracking faster and more practical.

Instead of typing meals, you:

take a photo
let AI identify the food
instantly get estimated nutrition data

This includes:

carbohydrates
calories
macros
Why this matters for diabetes

Carb awareness is critical.

But manually logging every meal creates friction — and that friction leads to inconsistency.

MetricSync reduces that friction by removing the need to:

search food databases
estimate portions manually
log everything by hand
How it works
Take a picture of your meal
AI analyzes the food
Get instant estimates for carbs and nutrition

This can help with:

quicker decision-making
better awareness of intake
more consistent tracking
Important note

This is a support tool, not a medical device.

It does not replace:

professional medical advice
glucose monitoring devices
prescribed treatment plans
Pricing
Free for 3 days
Then $5/month
Try it

https://metricsync.download/diabetic

Summary

For diabetes management, consistency matters more than perfection.

Reducing friction in food tracking makes consistency more achievable.

MetricSync — AI-powered nutrition tracking without manual logging

JackAltman — Sun, 22 Mar 2026 19:27:46 +0000

Tracking calories is annoying.

Typing in every meal, guessing portions, searching databases — most people quit because it’s too much friction.

MetricSync approaches this differently.

What is MetricSync?

MetricSync is an AI-powered nutrition tracker that lets you log food by simply taking a photo.

Instead of manual input, it uses AI to:

recognize food from images
estimate portion sizes
calculate calories and macros

No searching. No typing.

How it works
Take a photo of your meal
The AI identifies the foods
You instantly get:
calories
protein
carbs
fats

It works with:

homemade meals
restaurant food
meal prep
Why this is useful

Most calorie tracking apps rely on manual entry.

MetricSync removes that step entirely.

That matters because consistency is the hardest part of nutrition tracking — and reducing friction makes it easier to stick with.

Additional use cases

MetricSync isn’t just for general fitness.

It can also help with:

weight management
macro tracking
diabetes-related tracking and decision support ()
Pricing
Free for 3 days
Then $5/month
Try it

https://metricsync.download

Summary

Most nutrition apps optimize for data input.

MetricSync optimizes for speed and simplicity using AI.

I launched a tiny macOS tool yesterday and woke up to 100+ installs

JackAltman — Wed, 04 Mar 2026 06:00:05 +0000

Yesterday I shared a small tool I was working on with friends called TokenBar and the response honestly surprised me.

When I started building it, the goal was simple.

While working with the OpenAI and Claude APIs I kept running into the same problem during development: token usage is basically invisible while you're actually coding.

You run prompts.
You test agents.
You debug requests.

Everything works… but you have no quick way to see how many tokens you're burning until you check dashboards later.

After accidentally burning more tokens than expected during prompt experiments a few times, I wanted something extremely simple: a small indicator that shows token usage live while I'm developing.

So I built TokenBar, a macOS menu bar utility that shows token usage in real time when working with LLM APIs.

The idea is basically to treat token usage like a system metric — something you can glance at while you're coding.

No dashboards.
No heavy analytics tools.
Just a small counter sitting in the menu bar.

I shared it yesterday and ended up getting 100+ installs in the first day, which I definitely didn't expect.

A few developers have already said it's been useful for:

• prompt experimentation
• debugging LLM agents
• keeping an eye on token costs while coding

Still early, but it's been really interesting seeing how people are using it.

If anyone here builds with LLM APIs and wants to check it out:

https://tokenbar.site

Would also love feedback or ideas for improving it.

I accidentally burned $27 in LLM tokens during a single debugging session

JackAltman — Wed, 04 Mar 2026 00:01:42 +0000

I was debugging an LLM tool a few weeks ago and something weird happened.

Nothing broke.
No errors.
Responses looked normal.

But when I checked usage later, I had burned $27 in tokens during what I thought was a quick debugging session.

The reason was simple: I had no visibility into token usage while I was actually developing.

The typical workflow with LLM APIs looks like this:

write prompt
run request
check response
repeat

But token usage is basically invisible in that loop.

You usually only see it:

• later in dashboards
• inside logs
• after the fact in billing

Which means when you're experimenting, it’s very easy to burn tokens without realizing it.

Especially when you’re:

• testing prompts
• looping requests
• running agents
• debugging API calls

After running into this a few times I wanted something extremely simple:

a live token counter while I'm coding.

Something like how your Mac shows CPU usage or battery percentage.

So I built a small macOS menu bar tool that shows token usage in real time while you're working with OpenAI or Claude APIs.

No dashboards.
No analytics platform.
Just a small number in the menu bar showing how many tokens are being used.

It ended up being way more useful than I expected, especially during prompt experiments.

If anyone here builds with LLM APIs and has had similar “wait how many tokens did that use?” moments, you can check it out here:

https://tokenbar.site

Also curious how other people monitor token usage during development. Logs? Custom tooling? Something else?