I'm a solo founder. It took me 9 months and at least 3 stack rewrites to ship my SaaS.
I am a solo founder. Nine months ago I started building an SEO tool. Two weeks ago I got my first Stripe payment. In between, I rewrote the stack maybe three times and built things I had never touched before: a backend, a database, webhooks. All of it for the first time, with AI as my teacher.
I am not a developer and I am not a CSS person. I came in with 15 years of SEO experience, some project management, and a digital illustration hobby. Everything technical I figured out as I went.
The product is IvaBot - an SEO tool with three modules (technical audit, content coverage, content generation). Checks how ready a site is to be cited by AI engines like ChatGPT and Perplexity. Pay-as-you-go from $5, no subscription. https://ivabot.xyz
Everyone here ships in 2-3 months. I took 9!
What I tried first (and threw away)
Plan A: a Telegram chatbot. Realized about 10 people would actually use it.
Plan B: a Typebot chat embedded on a Webflow landing. Built the whole decision tree inside Typebot's free tier. That phase taught me how to write prompts, structure logic, make HTTP requests, parse JSON, set up webhooks, work with a real database. ChatGPT was my teacher for 6 months.
Stack: Webflow + Typebot + Memberstack (auth) + Stripe + Supabase + SerpDev + Make.
Monthly cost: ~$200/month before a single sale. Unsustainable.
A designer friend did a rough homepage draft for free, then I was on my own. Copied his tokens into Webflow, built everything else, drew the logo, made all the homepage illustrations by hand.
What changed: switching to Claude
Moved development work to Claude. Build sped up significantly.
Rebuilt the backend. Files to GitHub. Killed Typebot. Killed Memberstack (Supabase Auth instead). Swapped SerpDev for DataForSEO (PAYG, much cheaper).
New monthly cost: ~$40/month. Around 5x reduction.
Where it is now
Launched April 29, 2026 with live Stripe
First paying transaction two weeks ago
One blog post going slightly viral about how ChatGPT, Perplexity, Claude, and Google AI cite different sources for the same query
First impressions and clicks from Google showing in Search Console
Three modules, $5 starting price, pay-as-you-go. Built for solo founders and small businesses because every other tool in this space starts at $50-300/month, and I know the pain.
What I would do differently
GitHub repo on day one. Not a no-code stack.
Skip Memberstack. Spend that money on Claude credits and build auth properly.
Think about security from day one, not bolt it on later.
Use Claude earlier.
What's next
Product Hunt launch May 27. Next month: expanding Content Coverage with bigger automated checks for AI citation readiness, and a chat widget for the blog (grounded in site content).
Why I'm posting this
I do this alone. My partner is patient but tired of hearing about Stripe webhooks. I do not really have anyone to talk shop with.
Would appreciate honest feedback on the product, stories from other slow solo founders, or just a "yeah, this is hard", that works too.
Anyone else done a 6-9 month solo build? How did you balance dev time against marketing time?
Nine months is not slow — it is honest. The founders I have consulted with who shipped in 6 weeks usually spent the next 6 months untangling tech debt that blocked every feature they wanted to build next. You built something real and got a paying customer. That is the actual benchmark.
On dev vs marketing balance: strict time-boxing works better than splitting every day. A week of focused build followed by a week of focused outreach tends to produce better output than 4 hours of each daily. Context switching at this stage is genuinely expensive.
The stack cost reduction is underrated as a signal. Going from $200 to $40/month before revenue is the kind of infrastructure discipline that keeps you in the game long enough to find product-market fit. Most no-code-heavy stacks quietly become anchors around month 8-12 when you need to add something the tools do not support.
The AI citation readiness angle is real and differentiated. Curious: are you tracking which content attributes (structure, source links, entity density) correlate most with getting cited across different AI engines? That data would be genuinely valuable product signal.
Thanks for this, the 9 months framing especially! That's the part I needed to hear today, the constant "shipped in 6 weeks" stories are exhausting when you know what it actually takes solo.
The time-boxing point lands too. I had been splitting days and feeling the cost without naming it. Switching to week blocks.
On the correlations question: not yet at the scale to claim real patterns from my own data. Working from my own testing and 2026 benchmarks for now, which point to a consistent stack of factors (structure, schema, original data, depth, freshness, author signals). The next module I'm building is around exactly that, surfacing which of those are missing on a given page. The data will sharpen as more sites use it.
The approach of grounding in benchmark data first, then letting product usage sharpen the signal, is exactly how good data products get built — start with hypotheses, validate with real usage over time.
Of those six factors you mentioned, freshness and author signals are probably the most underweighted by traditional SEO tools. Most platforms still treat a page as a static object, but AI engines seem to reward "live" content and identifiable expertise behind the writing.
Good luck with the Product Hunt launch on May 27 — shipping solo through 9 months and 3 stack rewrites to get there is genuinely impressive work.
This is way more honest than most “I built a SaaS in 2 months” stories.
The part people always skip: the rebuilding phase.
3 stack rewrites in 9 months isn’t a failure — it’s literally what happens when you’re:
learning backend + auth + infra from scratch
validating product direction at the same time
and trying to keep costs alive long enough to survive
What actually matters here is this:
👉 you didn’t stop
👉 you simplified the stack
👉 you reduced costs from $200 → $40/month
👉 and you got your first Stripe payment
That’s the real milestone, not the timeline.
Also, the shift from no-code tools → Claude-assisted custom backend is a big inflection point. That’s usually where a project stops being “a prototype” and becomes “a real system.”
Most people never get past that middle stage where everything is messy and expensive.
Respect for shipping this solo — especially in a space that’s evolving as fast as SEO → AI search.
Curious: what ended up being the biggest time sink for you — auth, backend logic, or integrations like Stripe/webhooks?
good luck!
This felt very real to read.
The part about rebuilding the stack multiple times and learning backend/webhooks/database concepts while actively building resonated a lot with me. I’m also building solo right now, and I think people massively underestimate how mentally difficult the “unfinished middle” phase is — where you’ve spent months building but still don’t have clear validation yet.
Getting the first Stripe payment after all that probably felt incredible.
Also agree on the AI point. Tools like Claude and ChatGPT have genuinely lowered the barrier for technical execution, but they still don’t replace persistence, product judgment, or actually sticking with the problem long enough to ship.
Really respect the transparency in this post.
Congrats on shipping, Galyna! 9 months as a solo founder is incredibly honest work, especially doing stack rewrites along the way. Shifting from Memberstack to Supabase Auth to save on monthly runway is a massive win that most founders overlook early on.
I’m currently building Velaris, which is a curated marketplace specifically for Micro Tools, Dev Tools, and Datasets. We launched it because we noticed so many independent builders waste months trying to figure out distribution and marketing after spending a year building their tools.
On balancing dev vs marketing: Time-boxing works wonders. We spend one sprint purely focused on code/design, and the next sprint strictly doing SEO optimization and community outreach.
Your AI citation readiness module sounds highly differentiated—would love to see how you structure your data filters!
This is a strong founder story, but the real product angle is sharper than “SEO tool.” IvaBot is really sitting in the new AI visibility layer: helping small businesses understand whether ChatGPT, Perplexity, Claude, and Google AI can actually cite them.
That matters because traditional SEO tools still talk in rankings, keywords, and audits, while the buyer pain is shifting toward “will AI engines understand and reference my site?” That is a cleaner category than broad technical audit plus content generation.
The naming layer is where I’d be careful. IvaBot.xyz makes the product feel like a small chatbot experiment, even though the actual product is more serious: AI citation readiness, content coverage, and technical visibility. If this grows into a broader AI search intelligence platform, Beryxa .com would carry more trust and SaaS weight than a bot-style .xyz name.
Thanks for the feedback. To answer the naming question, the original idea was actually a chatbot, so the name came from there. And as I mentioned, budget was a constraint, good .com domains are not cheap. The .xyz was a practical choice at the time.
I get what you're saying about brand perception, and a future move to a different domain is not off the table. But for now I think if the tool is genuinely good, the audience will find it regardless of the TLD. We'll see how it plays out.
That makes sense. If the original wedge was a chatbot, IvaBot was a logical starting point.
The part I’d be careful with is assuming the audience will find it regardless of the TLD. That can be true once trust is already established, but early on the name and domain are part of the trust signal.
Especially in your category, small businesses are already skeptical of SEO tools, AI tools, audits, and “visibility” promises. Before they test the product, they’re judging whether it feels credible enough to try.
So the risk is not just .xyz versus .com. It’s that IvaBot still frames the product as a bot, while the stronger direction is AI citation readiness and search intelligence.
I also agree that many good .coms are expensive, but not every strong brandable .com has to be impossible for an early founder. Sometimes if the fit is right, there are founder-friendly ways to make a cleaner name work before the product gets too attached to the current one.
If this stays a lightweight chatbot-style tool, IvaBot.xyz is fine. But if the goal is a serious AI visibility platform, I’d pressure-test the broader brand before too much content, user memory, and search presence builds around the old chatbot frame.
Beryxa .com fits that broader SaaS direction better because it does not lock you into “bot” or “SEO tool.” It gives the product more room to become a real AI visibility layer.