Latest / The Indie Hacker Podcast with Fexingo: Solo Developers, SaaS Side Projects, and Independent Tech / How One Solo Dev Built a SaaS Using No-Code and AI
Transcript
- Lucas: If today's tech conversation gave you something usable, maybe a new tool or a fresh way to think about building, that's the whole point. But before we dive in — a quick honest thing. A handful of listeners chip in monthly through buy me a coffee dot com slash fexingo, and that's literally what funds making this many episodes. No ads, no sponsors, just people who find value here. It keeps the show independent, and we're grateful. Luna: Yeah, it's been that way from the start, and it lets us pick topics without worrying about what an advertiser might think. Alright, so today — no-code plus AI. That combo is getting a lot of hype, but we found a case that actually has real numbers behind it. Lucas: Right. So there's a solo developer, goes by the handle DevInTheCloud on Twitter — he's been sharing his build in public for about eight months now. He launched a SaaS called DocuMind — basically a document analysis tool for small law firms. Think: upload a contract, get a summary of key clauses, risk flags, and suggested edits. All powered by GPT-4, but built almost entirely with no-code tools. Luna: And when you say no-code, what does that actually mean in practice? Like, is it all drag and drop? Lucas: Pretty close. His stack is Bubble for the frontend and user management. Airtable as the backend database — storing user data, document metadata, analysis results. And then he uses the GPT-4 API through a middleware layer on Bubble's backend workflows. So he's writing zero traditional code — no Python, no JavaScript, no server setup. Bubble handles hosting, authentication, even payment processing via Stripe integration. Luna: So the only 'code' in the whole product is the API call to OpenAI. That's wild. And what did it cost him to build the first version? Lucas: He says about $300 and three weekends. Bubble has a free tier, Airtable's free plan is generous, and the GPT-4 API cost him maybe $50 in test credits. He built a landing page first with a mockup and a waitlist form, got 150 signups in two weeks, then built the MVP for those people. That validation step is key — he didn't build anything until he had proof people wanted it. Luna: That's the classic indie hacker move, right? Landing page first, product later. But I'm curious — once he launched, what happened? Did people actually pay? Lucas: They did. He launched with a single $29 per month plan. In the first month, he converted about 8 percent of his waitlist, hit $2,100 MRR. Six months in, he's at $8,000 MRR, growing about 15 percent month over month. Not hockey-stick, but steady. And he's still solo, still running the whole thing himself. Luna: That's impressive, but I have to ask — how does he handle support? If you're solo and you've got paying customers, support can eat your whole day. Lucas: He uses a combination of automated replies for common questions, a knowledge base built with Notion, and a simple ticketing system inside Bubble. He says support takes maybe two hours a week. The product is straightforward enough that most issues are about account setup or billing. The AI analysis itself rarely fails — but when it does, he's got a fallback that returns a 'Could not analyze, please try again' message and logs the error so he can fix the prompt. Luna: So the AI is the core feature, but the support burden is low because the scope is narrow. That makes sense. But here's what I'm wondering — does no-code create a moat? Or is it a commodity trap? Because anyone could build a similar thing with Bubble and GPT-4. Lucas: That's the big question. He's thought about it a lot. His answer is that the moat isn't the code — it's the data and the integrations. He's built integrations with Clio and MyCase, two popular practice management tools for law firms. Once a firm has connected their existing workflows to DocuMind, switching costs go up. Also, he's collected hundreds of anonymized legal documents that he uses to fine-tune his prompts, so the analysis gets better over time. That's hard to replicate. Luna: Integrations over AI — I love that. It's the same lesson from a lot of successful SaaS products: the features are copyable, but the ecosystem is sticky. So what's next for him? Is he planning to stay solo? Lucas: He's at a decision point. At $8,000 MRR, he's replacing a decent salary in many parts of the world, but he's in the US, so it's not quite full-time income yet. He's debating whether to raise prices, add a higher tier, or maybe hire a part-time support person. But he's resisted hiring because he likes the freedom of being solo. He's also considering whether to build a second product using the same no-code playbook. Luna: That raises an interesting question — is he an indie hacker if he's not writing code? The term traditionally means a solo developer, but if you're not developing, what are you? A product manager? A founder? Lucas: I think the label still fits. He's building a tech product, even if he's not writing the traditional code. He's designing, integrating, marketing, supporting — all the things a solo founder does. The tooling has just evolved. But I get your point — maybe we need a new term. No-code founder? Product-led solo operator? Luna: Product-led solo operator — that's a mouthful. But I think it matters because the indie hacker identity is tied to craft, the act of building. No-code changes the craft. It's more about configuration and integration than original creation. Lucas: And that's a trade-off. On one hand, you can ship fast and iterate quickly. On the other, you're limited by the platform's capabilities. Bubble has constraints — performance can be an issue at scale, and you can't customize the backend as deeply. But for a solo founder targeting a niche, it's often enough. Luna: I read somewhere that Bubble-hosted apps have a median load time of about three seconds, which is slower than custom code. But for a document analysis tool, that's probably fine. The real question is: can you build a million-dollar ARR business on Bubble? Lucas: There are examples. A company called AirDev built a seven-figure agency on Bubble. And there are a few solo founders doing six figures. But the ceiling is probably lower than custom code. That said, if you're aiming for a lifestyle business, not a unicorn, no-code is a legitimate path. DocuMind's founder is aiming for $20,000 MRR and then maybe a partial exit or just coast. Luna: Coast sounds nice. Let's talk about the AI angle specifically. He's using GPT-4 for document analysis — but how does he handle accuracy and hallucinations? Law firms can't afford errors in contract analysis. Lucas: Good question. He's not using raw GPT-4 output. He's built a structured prompt that asks the model to output JSON with specific fields: clause summary, risk level, suggested edit. He then runs a validation step that checks for required fields and sensible values. If the output doesn't pass, it retries or flags for manual review. He also shows a confidence score next to each analysis — so users know when to be cautious. And he's transparent that it's an AI assistant, not a replacement for a lawyer. Luna: That transparency is smart from both a legal and user trust perspective. So what's the biggest lesson from this story for someone listening who wants to build their own no-code SaaS? Lucas: I think it's three things. First, validate before you build — the landing page test is non-negotiable. Second, focus on integrations, not just features — that's your real moat. Third, embrace the constraints of no-code — they force you to keep the product simple, which is often better for a solo founder anyway. Don't try to build a platform. Build a tool that does one thing well. Luna: Well said. And on that note, I'm curious to check back in a year and see where DocuMind is. The no-code AI space is moving fast — maybe by then, the tools will be even more powerful. Lucas: Exactly. And if you're building something similar, share it with us. We love hearing about real experiments. For now, thanks for listening — and if you found this useful, consider chipping in at buy me a coffee dot com slash fexingo. It keeps the show ad-free and independent. Luna: See you next time.