Nick (nt)

Nick

ntik.me

9758 Followers

Recent casts

big milestone for me: I've learned 1000 Chinese words/sentences over the past 6 months. for the first 2 months, I used a bunch of off-the-shelf apps - all of which were in some way frustrating/annoying to use. so I built my own. Polylang is an app that I've been perfecting based on daily use. I've been using it on average 30 minutes every single day, with over 13k exercises completed just on my account. it's been helping me progress super fast. you import your vocabulary, automatically generate images and pronunciation audio, then work through flashcard exercises which are structured and timed automatically to improve retention and recall. once you get to a certain stage in practice, Polylang starts automatically remixing your vocabulary to serve unique generated sentences into daily practice. what's even cooler is that I own all of the data, and I built the app to satisfy the inner data nerd within me. I have full vocabulary tracking, with each item storing a model of how well my brain remembers it. I can view and track my progress in a very detailed way - it feels a lot like whoop or strava, but for language learning. I've offered in the past for friends to use it, but this is really the kind of thing that requires a serious daily commitment as it's pretty hardcore. but I've tuned it so that if you do show up every day, you basically end up learning 2-3x faster than the average learning, whilst spending less total time in practice. honestly it's worth vibe coding your own personal software. this probably took me 3-4 days of combined effort to build and refine, but it's absolutely perfect for me - and I don't really care if it doesn't suit anyone else but me. this cost me maybe $300-400 in tokens to build, but runs on maybe $2-3/month and I could probably optimise it to make it cheaper. so much better than spending $20/month on an app built for some target average demographic and facing the daily frustration of features that don't behave exactly how you want them to.

  • 1 reply
  • 0 recasts
  • 0 reactions

hot take: I actually liked the vibe of Farcon Rome so much specifically because it wasn’t hypey, shilly and salesey. felt so easy to skip all of that and just spend time connecting with people. livestreams could have been cool and I believe most of the talks were recorded, but I think everyone was too busy vibing IRL and spent less time casting about it i love that the main conference is a grassroots, community organised thing. it’s so easy to underestimate the amount of time organising effort it takes to put on any event of this size - especially where the attendees come from all around the world. really obvious how much love and care the organisers have put into Farcon Rome! that’s much more than I could have expected and asked for. had a blast. thanks!

  • 2 replies
  • 4 recasts
  • 43 reactions

Top casts

always keep coming back to the analogy of product building being so much like sculpting or painting. the first strokes are broad and confident. a lot of material gets applied and moved around very quickly. you might slap together 5-6 features in just hours and build this really huge thing out of nothing. but then the real work starts. you meticulously shape and form what you've created and refine the little details. subtle, barely noticeable changes. a lot of thought before each tiny improvement to a small part of the whole piece. the last 10% often takes 95%+ of the time. none of your users or competitors are even aware of what you're changing at this point. but it's what keeps all of your users coming back.

  • 57 replies
  • 18 recasts
  • 475 reactions

I'm building an app for language nerds and polyglots. I'm using it to learn Mandarin Chinese - it's helping me distill the insights from dozens of research papers on learning into a single, highly effective routine. I'm opening up access to a handful of alpha testers who are learning Chinese, German or Russian. Reply/DM if you're interested.

  • 115 replies
  • 11 recasts
  • 384 reactions

the trick to making voice agents fast is pipelining everything: first pipeline: audio packets -> speech to text -> turn-taking model second: LLM -> text to speech -> encoding -> output here's a render of my current latency. bear in mind, I'm running this locally from a wooden hut in the mountains in Turkey - it should be significantly faster once I host it!

  • 62 replies
  • 8 recasts
  • 391 reactions

Here's a mental map 👇 except it's not a map of some idea in your mind, it's a map of your mind and how well it remembered something. i'm using this to predict individual patterns of forgetting during learning in order to optimise how fast I can learn Mandarin.

  • 39 replies
  • 11 recasts
  • 334 reactions

Onchain profile

Ethereum addresses