I just published my 6 part blog series on my adventure into #gatsby and #netlifyCMS. It was super fun to work with these tools and hopefully others can learn from my experience! #LearnInPublic
1️⃣: https://t.co/HkAgeKniIv
2️⃣: https://t.co/zWv7BjPoUT
3️⃣: https://t.co/OfJCKbCUJV
@grok Regarding the AI Design tools: I tried all of them before I had the reference images and tweaked DESIGN.md as well. Most were horrible, Banani was decent.
But reference imgs + DESIGN.md + Banani = chef's kiss
Got a side project, but I suck at UX/design.
Finally found a working (AI) flow:
- @mobbin for finding designs I like -> store as imgs
- LLM to extract vibe from imgs & come up with DESIGN.md
- Feed DESIGN.md to @BananiAI for 1-shot design (really)
Really hope that Anthropic's internal cyber security is top notch now that they've public acknowledged they are sitting on thousands zero days / crit vuln... wouldn't want those to fall in the wrong hands. https://t.co/uGfRj58M4v
Took rslib for a first spin today, was super easy to set up a dual ESM/CJS output. Will deffo replace some of the other tsc based set ups.
I'm calling it today: Always bet on JS, always bet on Rstack tooling
@headless_baz@DannyDainton@getpostman Heya @headless_baz , I'm Postman engineer working on the update system, Danny seems to be out of office for longer period. Could you share details with me on Discord (PostmanPat), I'll deep dive to check what's going wrong!
This is what San Francisco does to you 😂
We designed a hypothetical 10km long 1km high zip line with a 3% incline and 30 mts segments.
Don’t tell me it won’t work, we will solve architectural issues after the first funding round 😂
@RussellCanfield Evals / testing, relevance of RAG (vs or combined w/ MCP & tool calling), sanitizing input & prevent jailbreaks, writing system prompts, splitting tasks for smaller models
@GergelyOrosz v0 for designs and rapid prototyping, which works really well as long as you let it pick its own technologies.
Then once a good base has been established, Cursor + Gemini 2.5 Pro to refactor into something that's maintainable enough to ship
"OpenAI lmnop6.5 ultra watermelon" 😂
"A mess of complex OOPisms, where you have to traverse 18 files just to follow a single behaviour" 🙈
"It’s trained on code that’s already an insult to silicon" 🔫
Funniest blog I've read in a while, can I sub to weekly rants, pretty plz?