You know what most SaaS products are designed to do? Grow past your actual needs.
Hootsuite promises “multi-platform scheduling.” You use it to post three times a week. Buffer offers “brand voice consistency.” You write in a voice that takes minutes to describe but hours to refine away from generic tools. Pixieset gives you gallery delivery and pricing tiers built for enterprise account managers.
The problem isn’t the tools. It’s that they’re designed to sell features you don’t need to the people who do.
For a blogger, a small creative team, a photographer running content alongside their actual work, enterprise tools are the wrong fit. They’re expensive, bloated with features designed for teams of ten, and they solve the wrong problem.
So what’s the right problem to solve?
The real bottleneck isn’t the platform. It’s the voice.
I spent four months using ChatGPT as a user. Copy something in. Wait. Refine it for hours. Post it somewhere.
The refining was the killer. ChatGPT gets you 70% of the way to your voice. Getting from 70% to 100% takes time.
I realized that if I could train a model once on my actual writing patterns, voice guidelines, word choices, tone, everything I never do, that last 30% would disappear. The refinement step would be unnecessary.
That’s not a SaaS feature. That’s a system design problem. And it only exists if you stay small and intentional.
How I built this (and why APIs are more accessible than you think)
I’m not a developer. I don’t code. But I knew I needed:
- A place to generate content in my voice
- A way to sync that to planning (Google Sheets)
- A way to publish to my blog (WordPress)
- Everything in one workflow, not five tools
That meant API integration. So I learned the math.
The token math that makes this work for small creators
Claude’s API costs are straightforward: you pay for tokens. A blog post is roughly 2,000-3,000 tokens. An Instagram caption is 200-400 tokens. A Pinterest description is 300-500 tokens.
At standard Claude pricing, a full blog post costs less than a penny. A month of my content (20 posts across formats) costs roughly 30 cents.
Compare that to:
- Hootsuite: €50-99/month
- Buffer: €40-99/month
- Pixieset: €9-24/month
- WordPress scheduling plugins: €15-50/month
A small creator needs maybe three of those. That’s €150-200/month minimum.
My API spend: 30 cents.
The math isn’t even close. But it only works because I built something intentionally small. I didn’t try to make it work for everyone. I made it work perfectly for me, and it turns out that same system works for anyone with a consistent voice and a workflow that needs integration.
Transferring your voice from ChatGPT to Claude
This is the part that matters technically. ChatGPT has no memory outside a conversation. Claude API lets you set a system prompt that lives with every request.
That’s where your voice lives.
Instead of describing my voice in a prompt (which doesn’t stick), I did one voice engineering pass:
I collected 10-15 examples of my actual writing: blog posts, email newsletters, social captions, everything.
I analyzed the patterns:
- Sentence structure (short, direct, occasional complex clauses)
- Words I never use (enterprise jargon, filler, em-dashes)
- What I always do (specific examples, avoid generalization, keep it honest)
- Tone signals (warm but not cheerful, helpful but not sales-y)
I wrote that as a system prompt, maybe 200-300 words total.
Now every piece of content generated through the tool starts with that voice already embedded. No refining needed.
This pattern scales because the architecture is the same everywhere
A newsletter writer could swap WordPress publishing for email delivery. A small marketing team could sync to Notion instead of Google Sheets. A technical writer could plug in their documentation platform instead of a blog.
The core stays identical: voice training once, integrate your actual workflow, eliminate refinement.
Which means the tool itself is sellable. Not as enterprise software. As something purpose-built for:
- Newsletters (voice consistency across your subscriber base without refinement)
- Bloggers (personality and depth without the 2-3 hours of editing per post)
- Small marketing teams (brand voice that scales across 2-5 people without hiring an editor)
- Technical writers (consistent tone and depth across documentation)
The tool is simple HTML and JavaScript. The magic is in the voice specification. That’s also the part that doesn’t scale as a one-size-fits-all product. Every creator’s voice is different.
So the product isn’t “a tool.” It’s “your tool, trained on your voice, integrated to your workflow.” Template-based, customized, lean.
Why constraints make better tools
SaaS products are designed to serve many people. That means they add features for edge cases. Features for enterprise use. Features for the day you grow big.
When you build for yourself, you delete everything that doesn’t solve your problem.
My tool does one thing: generates finished, publishable content in your voice across multiple formats. Blog, Instagram, Pinterest, carousel, email.
It doesn’t have:
- Trend analysis
- Engagement metrics
- Influencer suggestions
- Competitor tracking
- AI image generation
It does have:
- Voice training once, consistency always
- Direct integration to WordPress and Google Sheets
- A calendar view of scheduled content
- Minimal friction between idea and publication
The second list is smaller. But it’s the right size.
That’s the constraint that makes it work. Not for everyone. For the specific workflow of someone who:
- Writes consistently across multiple formats
- Needs voice consistency
- Works with their own CMS or WordPress
- Uses Google Sheets for planning
- Can’t afford enterprise tools
- Isn’t too small for an API
Who this applies to (and who it doesn’t)
This pattern works for:
- Photographers creating content alongside their main work
- Bloggers writing 2-4 times a week
- Small creative teams (2-5 people) with consistent voice
- Newsletter writers publishing across platforms
- Technical writers building tutorials or guides
It doesn’t work if:
- You’re hiring a content agency (they need different tools)
- You post daily across 10 platforms (you need enterprise scheduling)
- Your team doesn’t have shared voice guidelines
- You’re managing user-generated content
The specificity of what it solves is why it works.
What I learned from building my first API integration (with zero development background)
First lesson: understanding the problem is harder than implementing the solution. I spent more time defining what my voice actually was than building the integration.
Second lesson: APIs are less scary than they look. You don’t need a CS degree to send JSON to Claude and get text back. You need:
- Clear requirements
- Patience to read documentation
- A way to test (I used my own server)
- Willingness to iterate
Third lesson: the expensive part of content creation is refinement. If you can eliminate that step, the rest becomes fast and cheap.
The question
Are you paying for subscriptions because you genuinely need those features, or because there’s no other option available?
If it’s the latter, and your workflow is specific enough to describe clearly, building might be cheaper than subscribing.
And if you’re worried about the coding part: I didn’t know how to do this either. I just had a clear problem and time to iterate.