The problem Content creation was eating my time in the worst way: through friction, not volume. I wasn’t struggling to come up with ideas. I
Content creation was eating my time in the worst way: through friction, not volume.
I wasn’t struggling to come up with ideas. I was struggling with the tooling.
I started using ChatGPT to draft blog posts and social captions. It worked, but only partially. ChatGPT would generate something close to my voice, and then I’d spend 4-6 hours refining it. Changing word choices. Removing generic phrases. Adding specificity. Trying to make it sound like me instead of like an AI writing generic advice.
Meanwhile, I was juggling platforms. Write in ChatGPT. Copy to Google Sheets for planning. Manually post to Instagram. Schedule Pinterest separately. Draft the blog post in WordPress. Five different tools doing the same job: moving content from brain to audience.
And the costs added up. Hootsuite wanted a pro subscription just to unlock multi-platform posting. Buffer had the same limitation. SaaS pricing is designed to charge you more the moment you actually need to use the tool.
My budget was tight. I didn’t want to pay €50-100/month across multiple subscriptions just to post content consistently.
I realized the bottleneck wasn’t the platforms. It was the refinement.
ChatGPT got me 70% of the way to finished content. The last 30% required manual work to match my voice. And that 30% was taking most of the time.
What if I could remove that refinement step entirely? What if I had a tool that understood my voice from the start and generated content in my actual tone without hours of adjustment?
And what if I could do it all in one integrated system instead of copying between five different tools?
That’s when I started building.
I created a single HTML interface connected directly to Claude API, hosted on my own server. One tool. One place. Multiple content formats.
Here’s what it actually does:
Blog articles go directly to WordPress with a single click. Social captions sync to Google Sheets with metadata (platform, pillar, format, posting date, hashtags, engagement goal) so you can see your entire month at a glance. A visual calendar shows what’s scheduled and where the gaps are.
The magic happens in the voice training step. I trained Claude once on my actual writing patterns (word choice, structure, tone, what I never do, what I always do). Now every piece of generated content matches my brand voice from the first draft. No refinement needed.
Everything stays in one workflow: generate, schedule in the calendar, push to your CMS or social planning sheet, done. No copying between tools. No context switching.
Time saved:
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 saved:
SaaS subscriptions: €50-100/month across Hootsuite, Buffer, scheduling tools, and planning platforms
Custom tool: Claude API calls cost roughly 2 cents per post. Total spent: approximately 30 cents on a €5 credit
Result: Practically free compared to standard SaaS pricing
Voice consistency:
Before: Content shifted in tone as it bounced between tools and refinement cycles
Now: Same voice everywhere because everything comes from the same system, trained once
Result: Stronger brand identity, better audience recognition
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 ChatGPT alone wasn’t enough. I was stuck 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.
Then I realized I could automate the entire workflow, not just the writing step. I started building with Claude API. 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 computer science degree required. Just clear requirements and willingness to learn.
Building this taught me three things:
First, the expensive part of content isn’t the platform. It’s the refinement. ChatGPT got me 70% of the way. Custom voice training got me the last 30%. Once that step disappeared, the whole process became viable.
Second, constraint drives better design. I built this because I had a very specific problem: a photographer who needed fast, on-brand content without monthly subscriptions. That specificity is what makes it work. If I tried to make this for every photographer and marketer, it would bloat into a feature-heavy SaaS nobody needs. Instead, it does one thing perfectly: turns a photograph and brief description into finished, ready-to-publish content in my exact voice.
Third, you don’t need to be a developer to build custom tools anymore. You need clarity about your problem and patience to iterate. AI can help you understand the technical parts. You just have to be willing to learn as you go.
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
Let's talk about building a custom solution for your team.