Tried to solve my time constraints with an over-engineered AI automation. Now I have a digital twin who handles my demos, drinks my (digital) coffee, and has zero respect for the source code.
A fine line between a productivity hack and hostile takeover. 🧵
#BuildInPublic#AI
AI shows up and suddenly everyone wants to rebuild core tools in-house.
Meanwhile Anthropic runs on Zuora, Stripe, Workday, NetSuite, Salesforce and hires people to connect them.
The leverage comes from adding intelligence on top of proven systems.
A more interactive blog hero. Putting GPT-5.5 to good use solving the most pressing issues this weekend 😁 Inspo stolen from my friend @yeem_chantha
Step-by-step outlined here:
https://t.co/bud9xGOtlX
Tried to solve my time constraints with an over-engineered AI automation. Now I have a digital twin who handles my demos, drinks my (digital) coffee, and has zero respect for the source code.
A fine line between a productivity hack and hostile takeover. 🧵
#BuildInPublic#AI
He started getting a bit too comfortable with the brand. There’s a fine line between a helpful agent and imposter 🧵
#BuildInPublic#AI
Apparently, he thinks I'm just legacy code now.
He started getting a bit too comfortable with the brand. There’s a fine line between a helpful agent and imposter 🧵
#BuildInPublic#AI
Apparently, he thinks I'm just legacy code now.
@GrindChronicles @mscode07 Looks nice! What do you use as source for book data? OpenLibrary?
I built something similar a couple years ago, but before AI-assisted workflows so I reckon I could refactor/improve quite a lot if I get a weekend :)
https://t.co/oFIKztZHOf
@mscode07 Just a Sunday project from yesterday:
https://t.co/LoPNknNRRL
Stage 1 scans for tells (em dashes, throat-clearing, hype verbs) tuned for English prose. Stage 2 adds a two-sentence roast, generated on the server in English.
We’ve identified a security incident that involved unauthorized access to certain internal Vercel systems, impacting a limited subset of customers. Please see our security bulletin:
https://t.co/0S939n3qHC
I used the increased weekend quota of @cursor_ai in the new Agent Window to build a tool that hates AI writing. The irony is delicious.
Em Dash Patrol: https://t.co/FzWj8nika7
MVP in 1 hour. 2 hours fighting Cloudflare Turnstile 🤪
Try it out!
Built Em Dash Patrol, paste any draft and it flags classic “AI slop” tells (em dashes, throat-clearing, jargon), then Gemini gives a blunt editorial roast. Nuxt on Vercel.
https://t.co/0aqIEnH0qA
Mostly as an excuse to take @cursor_ai new Agent Window for a spin with double Composer 2 quota over the weekend.
And theeeere goes the Claude allowance for this Sunday 😄 Lollygagging its way towards the Pro limit. Switching to Composer 2 for now.
@cursor_ai offers double quota this weekend in their new Agents Window (not sure if only applies to tasks initiated from Agent window or also in terminals/side bar @amanrsanger@mntruell ?). Let's see what it can do.
The paralysis-by-analysis with all the opportunities is real! I like the idea of just carving out that time and going for a tool. 20 mins sounds like way too little though. Wiring a new tool into a proj, test it, iterate. That's 2 hours gone, easily. At least for me.
Today I'm trying Langfuse, as I haven't implemented any observability in my MCP projs up until now.
The AI dev paralysis-by-analysis is real... Should I finally build one of those 30 side projects I've got listed from the past few years, now that it's possible to realize them over a weekend, or re-factor the projects I built sloppily before AI?
https://t.co/ImlOWynwJz and https://t.co/pl6bD31oUa are both screaming for AI-assisted iterative improvements. They were built before AI tools and in little time (weekends and evenings).
Forks in the road, everywhere...