A few months ago I realized I was spending three full evenings every month planning content.
Not creating it. Planning it. Thinking through what to photograph, how to describe it for Instagram, where it would fit on Pinterest, how to write it for my blog so it felt consistent across all three.
And then another 4-6 hours refining the writing itself because the tools I was using weren’t capturing my voice.
The problem
I was using ChatGPT to draft content, then spending hours refining it to match how I actually write. Meanwhile, I was juggling scheduling across platforms. Hootsuite wanted a pro subscription just to unlock multi-platform posting. Buffer had the same limitation. Every tool had a pricing tier designed to make you pay more the moment you actually needed to use it.
My budget was tight. I didn’t want to pay €50-100/month across multiple subscriptions just to post content consistently.
The gallery delivery was handled separately (Pixieset for that). But for content creation and scheduling, I was stuck paying for features I barely used across multiple disconnected tools.
The insight
The bottleneck wasn’t the platforms. It was the refining. ChatGPT could generate something close to my voice, but it always needed adjustment. Hours of adjustment.
What if I built something that understood my voice from the start? Something that generated content in my tone without all the refinement work. One integrated workflow instead of five fragmented tools.
From ChatGPT to custom tool (and I’m not a coder)
I should be clear: I’m not a developer. I have no coding experience. I don’t know PHP, JavaScript, or any programming language.
But I knew I needed something better than just using ChatGPT directly as a user.
At first, I was in the AI Ladder phase. Copy and paste into ChatGPT, get output, refine it, move it somewhere else. It worked, but it was still manual. Still fragmented. I wasn’t saving time, I was just replacing one tool with another.
Then I realized I could automate the workflow itself, not just the writing. So I started building. Using AI to help me understand what I needed, asking questions instead of coding from scratch, testing things on my server, iterating until it worked.
No prior development experience. Just patience, clear requirements, and willingness to learn by doing.
The approach
I built a custom HTML tool hosted on my server, connected to Claude API. Upload a photo. Write a brief. Generate.
Claude understands my voice because I trained it once on my actual writing. It generates blog posts, Instagram captions, Pinterest descriptions, all in my exact tone. No hours of refinement needed.
Everything syncs to Google Sheets for planning. I publish directly to WordPress when ready. One integrated system, not five separate subscriptions.
What started as using ChatGPT as a user evolved into building infrastructure around it. No coding degree required. Just understanding my problem clearly enough to build the right solution.
Why this matters
When you’re using multiple tools and spending hours refining the output, you’re not creating. You’re managing.
The metrics are clear:
Time:
- Before: 3 evenings planning + 4-6 hours refining per content batch = 15-20 hours per month
- Now: 1 evening planning and creating = 3-4 hours per month
- Result: 75-80% time reduction per month. Over a year, that’s roughly 150+ hours freed up for actual photography and strategy.
Cost:
- Each API call costs 2 cents
- I’ve spent 30 cents so far on my €5 API credit
- Compare that to €50-100/month for Hootsuite, Buffer, or similar tools
- Result: practically free compared to standard SaaS subscriptions
More importantly, I got consistency. Same voice, same quality, everything from one source. No dilution through multiple platforms or tools.
The principle
Most SaaS tools charge you for the feature you actually need. Want multi-platform scheduling? Pro plan. Want brand voice consistency? Another tool, another subscription.
When you have a specific workflow and a tight budget, you stop shopping. You build.
What I learned
Building this taught me that the expensive part of content isn’t usually the platform. It’s the refinement. If you can remove that step, everything else becomes fast.
ChatGPT got me 70% of the way. Custom training on my actual voice got me the last 30%. And suddenly the whole process became viable.
The second lesson: constraint drives better design. I built this because I had a very specific problem. That specificity is what makes it work. If I tried to make this for every photographer or marketer, it would become a bloated SaaS product. Instead, it does one thing exceptionally well.
The question
Are you paying for subscriptions because you need the tool, or because it’s the only option available?
If it’s the latter, and you have a specific workflow, building might be cheaper than subscribing.