Forem: Team Awesome

AI's Hidden Environmental Cost: What Every Developer Should Know

Team Awesome — Sun, 18 Jan 2026 23:37:01 +0000

My daughter asked me something the other day that I couldn't shake: "Dad, am I hurting the environment every time I use ChatGPT?"

I didn't have a good answer. So I spent a week digging into the research. Here's what I found.

The Numbers Are Wild

A single ChatGPT query uses about 0.3 watt-hours of electricity. That's 10x more than a Google search. Sounds small until you remember there are over a billion AI queries happening daily.

But electricity is just the start. Every prompt you send requires cooling. A single 100-word prompt uses roughly 500ml of water when you factor in data center cooling. That's two cups of water. Per prompt.

Large data centers gulp 3-5 million gallons daily. That's 5-8 Olympic swimming pools worth of water. Every. Single. Day. And 80% of it just evaporates.

The Data Center Jobs Myth

This one surprised me the most. You've seen the commercials about AI data centers bringing jobs to middle America. The reality?

The entire United States has only 23,000 permanent data center jobs. That's 0.01% of employment while consuming over 4% of electricity.

OpenAI's Stargate project in Texas? 1,500 construction workers, then 100 permanent jobs. Taxpayers subsidize these positions at an average of $1.95 million per job.

Virginia's own auditor found the state generates only 48 cents in economic benefit per dollar of tax incentive. Net loss.

What Developers Can Actually Do

The good news: your choices matter more than you'd think. The right strategies can cut your AI footprint by 50-90%.

Model selection is everything. An 8B parameter model uses 60x less energy than a 405B model. Don't use Claude Opus for tasks Haiku can handle.

Prompt engineering saves more than you'd expect. Trimming verbose instructions and unnecessary context can reduce token usage by 30-50%. One company dropped from $5,000/month to $1,500 just by optimizing prompts.

Caching is massively underutilized. Both Anthropic and OpenAI offer prompt caching where cached tokens cost only 10% of regular tokens. If you're sending the same system prompt repeatedly, you're wasting 90% of that energy.

Context windows add up fast. AI doesn't remember your conversation. Every message resends the entire history. A 50-message chat means re-reading 49 messages before responding to the 50th. Start fresh when switching topics.

The Full Breakdown

I wrote up the complete research with specific model comparisons, efficiency tiers, water consumption data, and a breakdown of why price doesn't always correlate with energy use (spoiler: reasoning models like o1 use 50-100x more compute despite similar pricing).

Read the full post: AI Energy Consumption: How Much Power Does AI Really Use?

Also, if you want to see how many tokens you're actually sending before hitting that API, I built a free tokenizer tool that supports GPT-4, Claude, Gemini, and others.

What's your take? Are you factoring energy consumption into your model choices, or is it not even on your radar yet?

I Applied for AdSense and Got Rejected for "Low Value Content"

Team Awesome — Tue, 30 Dec 2025 22:05:56 +0000

A month ago I posted here about building a developer tools site using AI. Since then I've added more tools, written blog posts, built out an API directory.

I figured it was time to monetize. Applied for AdSense. Got rejected.

The reason? "Low value content."

Wait, What?

Every tool page has explanations, FAQs, use cases, step-by-step instructions. I have an about page, privacy policy, contact page. Original blog posts. This isn't a thin site with an input box and a wall of ads.

But Google disagreed.

What I Learned

After researching this, "low value content" is basically Google's catch-all rejection for new sites. The actual reasons have little to do with content quality:

Site age - Google wants 3-6 months of history before they trust you. My site is too new.

Traffic - They want sites that already have visitors. Classic chicken-and-egg problem.

Tool sites get flagged - Too many spammy utility sites have ruined it for everyone. Google's algorithms are skeptical by default.

What I'm Doing About It

Not panicking. The content is solid. My plan:

Wait 3-4 weeks and reapply. Many people get approved on attempt #2 or #3 with zero changes.
Keep publishing. More content, consistent activity.
Focus on traffic first. SEO, community engagement, being actually helpful in places like this.
Consider alternatives. Ezoic, Carbon Ads, or just keep building and worry about monetization later.

The Real Lesson

Chasing AdSense approval is probably the wrong priority for a new site. The goal isn't to run an ad farm. It's to build something useful.

If I do that well, monetization will follow. And by the time Google approves me, I probably won't need them as badly anyway.

I wrote a longer breakdown on my blog if you want the full story: What "Low Value Content" Actually Means

Anyone else dealt with AdSense rejection? What worked for you?

Here are some things I discovered trying to find the a decent AI tool

Team Awesome — Sun, 21 Dec 2025 06:37:31 +0000

I've been trying out a bunch of AI coding tools lately and I'm sure I'll be trying our a bunch more going into 2026. Started with Claude, Copilot, Windsurf, then Cursor, and somewhere along the way I realized I'd tested like a dozen of these things trying to figure out which ones actually help versus which ones I had to babysit to get them to cooperate.

Turns out they're all pretty different from each other. Some are autocomplete with extra features. Others try to be more like a coding partner or at least a junior programmer. A few let you describe what you want and they try to build it, which sometimes works and sometimes goes sideways.

The ones I actually use:

Cursor is what I use most days. The codebase chat is useful for understanding how old code works. Agent mode is inconsistent but when it works it can save you serious time. $20/month feels like a lot until you see how much faster things get.

GitHub Copilot is solid. $10/month, works in most editors, does what it says. Not as flashy as Cursor but it's reliable. If you have GitHub Pro you might already have access.

Some I found interesting:

Windsurf is free if you bring your own API keys, which is pretty smart. Being newer means less help online when stuff breaks, but the actual tool works well.

Aider is a terminal tool that works with any editor. Free and open source. Really good at making changes across multiple files. No autocomplete though, so it's more for bigger refactoring tasks. It's also more involved to get it up an running.

Claude Code (VS Code extension) is what I would use for contained objects, not great at planning a large project. Also, not great at autocomplete but the reasoning is solid. The $20 plan gets you a lot and resets every 5 hours or so.

Others I tried:

There are more. Amazon Q is AWS-focused. Supermaven is fast. Lovable and Bolt are for building apps without code. They all do different things, which is why asking "what's the best" doesn't really work.

Things I figured out:

The free options (Aider, Cline, Windsurf with your keys) are actually good. The paid ones are more polished and better at autocomplete but it's not a huge gap.

Being specific helps. Don't just drop an error in there. Give context, say what you tried, explain what you're working on. More info means better output.

Big projects with lots of dependencies still confuse these tools, like I was saying with Claude. They handle focused tasks well but get lost when too much is happening at once. Cursor does better with complexity, especially using the planning agent, but you still need to guide it.

If you want to try one:

GitHub Copilot if you're paying. $10/month, stable, lots of resources.

Cursor gives you a free week and you can use any model on an unlimited basis (or at least it feels unlimited), but once the week is over you best hit the breaks on the expensive models. If you don't you'll be throttled back or even hit the limit, although I never have.

Windsurf or Continue if you want free. You need your own API keys but then there's no limit.

Just pick one and use it for a week. You'll know pretty quick if it fits how you work.

Full comparison here:

I put together a breakdown of 18 tools with scores, pricing, what they're good at: Best AI Coding Tools Comparison 2026

Hope this helps, but please let me know what's working for you.

From Burnout to Builder: How AI Tools Changed My Relationship with Code

Team Awesome — Thu, 04 Dec 2025 22:49:19 +0000

I'll be honest, six months ago I was dreading opening my IDE. Every project felt like climbing a mountain in lead boots. The joy was gone.

I'm a data engineer by day, managing complex pipelines and infrastructure at a healthcare ad-tech company. I've always had side project ideas, but the gap between "cool idea" and "actually building it" felt insurmountable. Learning new frameworks, debugging for hours, fighting with CSS, it all felt like work on top of work.

Then I tried something different.

The Experiment

I'd been skeptical of AI coding tools. Felt like cheating somehow. Like I should be able to build things "the right way," learning every framework deeply, understanding every line of code I wrote.

But burnout doesn't care about your principles.

So I grabbed Cursor IDE and Claude, picked a simple idea (a collection of developer tools), and just started building. No grand plan. No trying to learn React from scratch first. Just building.

What Actually Happened

Here's the weird part: it worked. Really well.

I built Toolpod.dev, 48 browser-based developer tools like a JSON formatter, regex tester, base64 encoder, and a bunch more. Plus an API directory. Full responsive design, deployed on Firebase. The kind of project that would've taken me months of grinding through tutorials and Stack Overflow.

Did it in weeks.

But more importantly, I enjoyed it.

The Mental Shift

The fraud feeling was real at first. "Am I even a real developer if AI wrote most of this?"

Then I realized something. I've been using tools to be more productive my entire career. Frameworks instead of vanilla JS. Libraries instead of reinventing the wheel. Stack Overflow when I'm stuck. Git to manage versions. VS Code extensions for productivity.

AI is just another tool. A really powerful one.

What I actually do is architect the structure, make product decisions, review and modify AI-generated code, debug when things break (they do), optimize and refactor, and make it actually usable.

The AI handles the tedious stuff. Boilerplate, syntax I'd have to look up anyway, converting designs to code, repetitive patterns.

The Unexpected Benefits

The speed is obvious, but what surprised me is I'm not cutting corners. The code is solid. It's just faster.

I actually learn faster too because I can ask "why did you do it this way?" and get explanations in context.

Starting a new project doesn't feel daunting anymore. That's huge for side projects. The activation energy is so much lower.

And honestly, the joy is back. I'm building things again. Lots of things. The creative part is fun again.

What It's Not

Let's be real, it's not magic. You still need to understand what you're building. You'll hit walls and need to problem-solve. Bad prompts get you bad code. You're still responsible for what ships. Complex architecture still requires real expertise.

But for MVPs? Side projects? Learning new tech? It's a game-changer.

The Bigger Picture

I see posts about developers worried AI will replace them. I get it. But I think we're looking at it wrong.

AI isn't replacing developers. It's removing the grunt work that was burning us out in the first place. The tedious stuff that made us dread opening the IDE.

What's left is the interesting part. Solving problems, making decisions, building things people actually use.

Try It Yourself

If you're feeling burned out, or if you have ideas you've been putting off because they feel like too much work, just pick one small thing to build. Use Cursor or GitHub Copilot (both have free trials). Don't overthink it, just start. See how it feels.

You might surprise yourself.

For me, it changed everything. I went from dreading code to shipping projects again. The tools didn't make me less of a developer. They made me a happier one.

And honestly? That matters more than I thought it did.

I Built 48 Developer Tools in a Day Using AI - Here's What I Learned

Team Awesome — Sun, 30 Nov 2025 16:25:30 +0000

I wanted to build a side project that:

Costs almost nothing to run
Requires minimal maintenance
Could eventually generate passive income
Actually provides value to people

A few weeks ago, I launched Toolpod — a collection of 48 browser-based developer tools plus a directory of 110+ free public APIs. Here's how I built it using AI-assisted development, and what I learned along the way.

The Idea

Developer utility sites have been around forever. JSON formatters, Base64 encoders, UUID generators — we all use them. But most are covered in ads, require sign-ups, or send your data to a server.

I figured I could build a cleaner version:

All tools run client-side (your code never leaves your browser)
No sign-ups
Minimal ads (once AdSense approves)
Multiple tools = multiple SEO entry points = organic traffic over time

The Stack

I kept it simple and cheap:

Next.js 14 with static export
Tailwind CSS for styling
Firebase Hosting (free tier — 10GB storage, 360MB/day bandwidth)
Domain: toolpod.dev via Porkbun (~$12/year)

Total monthly cost: ~$21 (Cursor subscription + domain amortized)

Since everything is statically exported, there's no server to maintain. Firebase's free tier is more than enough for a utility site.

The Method: Cursor + Claude

Here's where it gets interesting. I didn't write most of this code manually.

My workflow:

Describe what I want to Claude in detail
Ask it to generate the code
Paste into Cursor or let Cursor's AI assist
Test, tweak, repeat

For example, here's roughly what I told Claude for the hash generator tools:

"Build a SHA256 generator page with these features:

Input text area

Encoding selector (UTF-8, UTF-16, Hex, Base64)

Output format selector (hex lowercase, hex uppercase, base64)

HMAC support with secret key input

Use Web Crypto API

Match the styling of my existing pages"

And it just... worked. The AI understood the context, generated the component, and I had a working tool in minutes.

What I Built

48 tools across categories:

Formatters: JSON, SQL, CSS, JavaScript, XML, YAML
Encoders/Decoders: Base64, URL, HTML entities, Base32
Hash Generators: SHA256, MD5, SHA3 with full encoding options + HMAC
Converters: JSON↔YAML, JSON↔XML, CSV↔JSON, TOML↔JSON
Generators: UUID, QR codes, passwords, cron expressions
Text Tools: Diff checker, word counter, case converter, slug generator
And more: JWT decoder, regex tester, color tools, chmod calculator

Plus an API directory:

110+ free public APIs organized by category (finance, weather, sports, AI, etc.). Each entry shows auth requirements, CORS support, and links to docs.

What Worked

1. AI-assisted development is a multiplier

I'm a data engineer by trade, not a frontend developer. Without AI assistance, this would have taken weeks. With it, I had a working site in a weekend and spent the following days polishing.

2. Static sites are underrated

No servers to maintain. No databases to back up. No costs that scale with traffic. For a utility site, this is perfect.

3. One tool = one SEO opportunity

Each tool page targets specific search terms ("json formatter online", "sha256 generator", "uuid generator"). Over time, these compound. I'm playing the long game.

What Didn't Work

1. Reddit hates new accounts

My first few posts got auto-removed by spam filters. If you're planning to promote on Reddit, build some karma first by answering questions before posting links.

2. AdSense takes forever

Still waiting on approval. Applied weeks ago. The site has all the "right" things (privacy policy, about page, real content), but Google moves slow.

3. Some tools are harder than expected

The basic stuff (JSON formatter, Base64) took minutes. But tools with edge cases (cron expression builder, JWT decoder with all algorithms) required more iteration.

The Numbers

48 tools built
110+ APIs in the directory
~$21/month total cost
$0 in revenue (so far — waiting on AdSense)
~1 weekend for initial build, plus a week of polishing

What I'd Do Differently

Start with fewer tools. I could have launched with 10-15 tools and added more over time. Shipping fast > shipping complete.

Set up analytics earlier. I added Google Analytics late and missed early traffic data.

Write content from day one. Each tool page has educational content now ("What is SHA256?", "Common use cases"), but I added it later. Should have been there from the start for SEO.

What's Next

Wait for AdSense approval
Monitor Google Search Console for indexing
Keep building karma on Reddit/DEV.to
Add more tools based on what people actually search for
Maybe open-source it?

Try It Out

If you want to check it out: toolpod.dev

If you're thinking about building something similar, my advice: just start. The AI tools available today make it possible to ship things that would have taken 10x longer a few years ago.

What side projects are you working on? I'd love to hear about them in the comments.