Before: My custom LLM wiki setup consisted of three prompts scattered across GitHub gists.
Now: it's a fully-fledged Claude Code + Codex plugin.
The llm-wiki plugin deploys a project wiki, adds wiki-aware research and planning skills, and makes the entire setup installable from the marketplace instead of copy-pasting bootstrap prompts and skills.
Solo dev here, been generating a ton of videos with Wan 2.2 the past few months for my side-project.
Needed somewhere cheap to dump thousands of clips + checkpoints.
Found Rabata while looking for S3-compatible storage, filled 3-field grant form, got approved for $100k credits the next morning. Zero cost since day one, uploads are fast, zero changes needed in code beyond the endpoint URL.
"If you can use Excel spreadsheet or Google form to build your product. Do it for your first 100 users." I did, what's next?
You need to have a MVP in the target market. You need to demonstrate that it can scale. Therefore, if your business model works for 100 customers on the web, it does not necessarily mean that it will be successful on a larger scale on mobile.
Technology is becoming increasingly complex. Customer expectations from the products are rising. 10 years ago, I single-handedly coded all the web components for a brokerage startup. Now, it seems impossible, perhaps only achievable with no-code tools.
It took me four years, starting with personal therapy and continuing with education in psychology and CBT, to switch from fintech to health tech. So no, it's not something you can do immediately. I also have an experienced psychologist on board.
I'm more like a crafter, so no code tools have provided me with the UX quality I was aiming for. But I was naive, thinking that experience would be enough. If I were to start now, I would definitely choose a simpler stack.
AI is pretty helpful right now (I'm proofreading my comments with Notion AI as we speak), but I think it will take a couple of years to really deliver quality output. I'm eagerly awaiting this moment.
There are two views on the point. One is that we should stick to traditional in-person therapy, use pen and paper, and be strict. The other is that we should help people wherever they are. If they use mobile devices, let's try to help them there.
"While there are always exceptions, you typically don't need to hire three developers to build a prototype.”
However, prototypes often don't show any traction. In addition to developers, you need marketing, design, and other people. Thus, three developers is just a simplification.
My main point is that the current market is highly competitive, and products are becoming increasingly complex, so more money is needed to reach our first customers.