Yes I too have found newer models (mostly Opus) to be much better at iterative development. With that being said if I have very strong architectural/developmental steer on what I believe the output should be [mostly for production code where I thoroughly review absolute everything] it’s better to have a documented spec with everything covered rather than trying to clean up via an agent conversation. In the team I’m in we keep all plan.mds for a feature, previously before AI tooling we created/revised these plans in Confluence, so to some degree reworking the plan is more an artefact of the previous process and not necessarily a best practice I don’t think.
First you have to be very specific with what you mean by idiomatic code - what’s idiomatic for you is not idiomatic for an LLM. Personally I would approach it like this:
1) Thoroughly define step-by-step what you deem to be the code convention/style you want to adhere to and steps on how you (it) should approach the task. Do not reference entire files like “produce it like this file”, it’s too broad. The document should include simple small examples of “Good” and “Bad” idiomatic code as you deem it. The smaller the initial step-by-step guide and code conventions the better, context is king with LLMs and you need to give it just enough context to work with but not enough it causes confusion.
2) Feed it to Opus 4.5 in planning mode and ask it to follow up with any questions or gaps and have it produce a final implementation plan.md. Review this, tweak it, remove any fluff and get it down to bare bones.
3) Run the plan.md through a fresh Agentic session and see what the output is like. Where it’s not quite correct add those clarifications and guardrails into the original plan.md and go again with step 3.
What I absolutely would NOT do is ask for fixes or changes if it does not one-shot it after the first go. I would revise plan.md to get it into a state where it gets you 99% of the way there in the first go and just do final cleanup by hand. You will bang your head against the wall attempting to guide it like you would a junior developer (at least for something like this).
Even if you don’t use tailwind the tailwind UI page has some nice example options. It’s nice in that they’re all on a single page so can quickly compare what UX might work for your use case.
Even outside of AI coding I have found a tremendous amount of value in using AI to produce a requirements and spec document for me to code from. The key unlock for me is asking AI to “interview” me about how this system/feature should work. As part of that process it will often ask a question that gets me thinking about interesting edge cases.
I will say I always provide an initial context document about the feature/system, to avoid us starting with trivial questions. After about 45minutes I’ll often feel I’ve covered enough ground and given the problem enough thought to really put pen to paper. Off the back of this I’ll ask it to summarise the spec and produce a document. This can be a good point to ditch AI if you are so inclined but still get value from it.
Beautiful. You can just tell when things are made with the core premise of how they wanted someone to FEEL playing it. Not some lore, not some "cool" gameplay ideas they had, not some fancy gfx concept they'd prototyped. It just feels different when you know it's been approached from a completely emotional perspective.
Just watched your video, really love the style and openness.
My only suggestion is niche it down a bit. The SQL tutorial guides and features sound great, but the functional list feels a bit like a laundry list. Even here you describe it as a tool for "developers/founders/teams".
Try targeting a specific domain, tech stack, database type, or developer segment (e.g., large B2B teams, small B2B teams, indie devs, or funded startup founders) to stand out. If you pick a clear niche, you can build a stronger SEO strategy around long tail keywords and tailor both the product and the messaging and work out what order to build out features. Even if long term you plan on wanting it to be a tool for all databases, segments etc.
It's much easier to produce content with this in mind, e.g. if you were targetting getting the most out of Postgres you could easily produce a bunch of content for PostgreSQL 18 which formally came out of beta a few weeks ago and has native support for UUIDv7 etc.
Fwiw I’m doing a ton with SQLite atm as a solo dev. If your landing page had said "THE VERY BEST TOOL FOR SQLITE MANAGEMENT TO HELP SOLO DEVS AND SMALL TEAMS MAXIMIZE SQLITE PERFORMANCE AND PRODUCTIVITY" there’s a good chance I would have signed up for updates but atm it felt a little generic, some of the features I might use, some I definitely would not.
True. 99% won't care, but that shouldn't hold you back. You get outsized returns from even a handful of people caring - feedback, amplification, motivation, moral support etc.
> Its just another form of marketing and it should be illegal to go around spamming posts advertising a product
I think this is the dogma that holds a lot of devs back, the belief that sharing your work, the product, the thought process, the journey, the mistakes, the wins etc is “spammy”. Would save your rhetoric for those who actually spam - ai slop generators, bots, link farmers, paid shillers etc. Not indie devs on HN trying to build something for the world.
I get the same whenever I get my daily walk in. Pure unbounded epiphany of ideas and experiments, surging with creativity. I'll revisit them a few days later and for 90% of them my immediate internal response is "that sounds like a really sh*t idea".
As a solo/indie dev who's currently early in building a product, I've been keeping a journal of "ideas" for content in a txt file in the codebase as I hate context switching and want to build this up before I get to it.
Here's what I've done:
- At the top of the file I've listed my audience, 3 personas
- My content has to be useful to one of those
- If I see an interesting post/take on social media I hold the link and write an idea for my own spin/take (takes 30 seconds) - log it
- If I have a problem/issue that I resolve that would be useful to my audience - log it
- If I have a key product/design/UX choice that took some time to think through - log it
- If something takes me much longer than I thought because there's more to it (iceberge effect) - log it
I've been doing this for about 6 weeks now and I've got 100 ideas for pieces of content.
One of the best pieces of advice I read is that when you're solo, many times people/community rally around you. You are the product too so you have to share what you're doing, it's interesting to many, not just your customers. They care about the advice you give, the input you have, the way you build things. You are a subject matter expert in this domain, so you should structure your content with this in mind.
"You escape competition through authenticity." - @naval
I get an immense amount of pride when I see anything that the Royal Museum Greenwich do because not only is it very local to me but they have such a great way of balancing the history of maritime/meridian with modern services and facilities.
If you’re coming to London please spend a day in Greenwich, you won’t regret it, take in the museums, the small markets, the observatory, stroll round the park, grab a photo of the best view in London (imo) https://www.rmg.co.uk/royal-observatory/attractions/enjoy-be... - then take a walk along the Thames, once you’re done you can hop on an Uber boat and head into central London.
I find the wordsmithery on Meta's statement the most interesting:
“We do not track your *PRECISE* location, we don’t keep logs of who everyone is messaging and we do not track the *PERSONAL* messages people are sending one another," it added. “We do not provide *BULK* information to any government.”
Even though it was a long time ago I still have IE6/7 workarounds burned into my brain, most of them float related but also having a whole stylesheet for that damn browser... <!--[if lte IE 6]>:
I remember working for a client who needed to support IE6 (with all the insane bugs/quirks/limitations) and I’d despair every time the designers would hand over a Photoshop design with rounded corners. They also needed it to be responsive (at the time mostly just different desktop sizes). Would usually require cutting the corners out and positioning them in table cells. There’s a certain amount of dev resilience you build having to do stuff like that by hand!
I have noticed a few people grumbling that the slow requests for Gemini in particular are taking upwards of 4-5 minutes to complete. I have noticed a massive decline in the slow request responses for Gemini in the last 48 hours too.