A bit more about sail:
The core idea is an inversion on agent coding orchestration:
- before: agent reads/writes files, runs tests. unless guided, over time architecture drifts, context gets bloated, it forgets to do stuff instructed.
- with sail: agent reads/writes code directly through the sail CLI. if it makes an a...
1
8
On the introspection debate:
- you can't get much done if you're constantly distracted by ego thought: "can I do this?" "will I actually achieve what I want" "wow I'm so great/powerful/terrible/strong/weak". the more you think about how what you're doing relates to your own conception of "I", the less time you have to...
2
14
having a lot of fun with my BitClaw. I keep giving it new tools and tinkering with it's source code to modify its core behavior. some things it does for me, now on a regular basis:
tracks workouts, gives nutrition advice
manages my inbox, auto-archives spam
schedules meetings, manages my calendar
feeds granola transcr...
5
2
23
built my own claw
- 1.2k LOC (vs. openclaw's 400k and nanoclaw's 7k)
- runs claude agent sdk in a container, so it's secure
- all the core stuff working: memory, crons and heartbeat
- it knows how to install and customise itself (just run claude in the dir)
- talk to it on telegram, or just ask it to upgrade itself to...
60
1
200
here's the app I've built to help me learn 60 new Chinese words every week. included: generate rich media flashcards, spaced repetition, AI sentence translation and a lot of nerdy stats (including a model of your memory!)
slowly growing the beta - we have 4 active users now! I'm looking to onboard a few more. If you'r...
18
1
92
Mandarin progress update: I am 14% to B1 (conversational) - about 1 year away at the current pace.
- I advance around 1.5% every week.
- learned 97 core words last month
- ~200 phrases/sentences learned last month
2x tutor sessions per week and daily flashcards + AI sentence translation exercises (roughly 100-150 per day)
35
47
300
my new recipe for learning and satisfying curiosities:
if I see a really cool piece of tech, I'll try to build my own version (without looking at the source code). even for complicated stuff, it's now possible to figure things out with one weekend and a bunch of LLM credits.
that's how you get deep down the stack and...
20
45
288
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 shoul...
62
9
391
okay a few hours later and I have performance comparable to that of Vapi/Elevenlabs agent SDK, albeit along the green path - I'm sure there are hundreds of edge cases that these companies spend much time figuring out, not to mention making their offerings flexible/observable etc
but it's crazy how quickly you can get ...
day 1 learning to build voice agent infra from scratch:
put together a VAD, twilio, deepgram (SST), o4-min, elevenlabs (TTS) into an event-based loop that coordinates listening and speaking
main issue right now is latency and quality of turn-taking/interruptions. not quite as good as Vapi/Elevenlabs off-the-shelf
ne
17
5
122
day 1 learning to build voice agent infra from scratch:
put together a VAD, twilio, deepgram (SST), o4-min, elevenlabs (TTS) into an event-based loop that coordinates listening and speaking
main issue right now is latency and quality of turn-taking/interruptions. not quite as good as Vapi/Elevenlabs off-the-shelf
ne...
23
24
222
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 th...
57
18
498
my Mandarin tutor is awesome. every session, he introduces ~25-35 new words and phrases, which builds my active vocabulary by ~20 words. I add them to my learning system, entering them into practice and consolidation - all managed by an algorithm that models my memory and manages my practice sessions.
within my syste...
68
14
282
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. Repl...
115
14
384
happy new year!
In 2026 I’m really excited to keep exploring the cutting edge of AI whilst looking for a non-software domain to dive deeper into.
I spent the last 6 months diving deep into supply chain with a good friend. whilst our interests eventually diverged, it was really exciting to explore a new domain that’...
15
27
175
been thinking about the Interslavic synthesised language
The slavic language family actually had 3 distinct sub-groups: eastern, western and southern.
Within the groups, speakers can mostly understand each other (70-90%) but outside the groups it’s impossible.
For example: I speak Russian and get the gist of Ukra...
1
6
Building a simulation of the one of the biggest problems in supply chains worldwide: the bullwhip effect:
A supply chain is the chain of companies that produces all of the things you buy every day. Every company in the chain takes some raw materials from upstream, changes/combines them into a product, and sells downs...
11
so interesting to learn a non indo-european language
I grew up speaking Russian, then learned English and more recently some German. though there are crazy grammar differences and obviously quite different vocabulary, if you're only ever exposed to indo-european languages you can take a bunch of stuff for granted.
f...
5
1
22
started building my own version control system over morning coffee
exploring these curiosities:
- how could version control look when buildings hundreds of toy projects with AI - perhaps branching from boilerplates, etc
- focus on single-user, removing as much complexity from the UX as possible
- maybe: compatible plu...
4
16
crypto's even deeper secret is that 80-90% of the farmers (users) come from Indonesia/The Philippines, who are there purely for the acquisition incentive so nobody is really ever getting acquired
the winning teams show the combined graph but never the breakdown by geography 🫡 💀
2
1
10
I find it difficult to understand the dogmatic simplification that exists in every single marxist take - it's always the exact same argument/narrative: company owners are stealing from workers and that's how they get rich and you don't.
I feel like it's easy to believe this dogma if:
- your priors are biased with the...
4
7
every tweet I read on my tl is LLM written. that’s not necessarily bad - the useful information is still there. but it feels… standardised, stylistically undifferentiated.
there is no-longer surprising individual craftsmanship, but efficient mass-production. human writing and thinking is on the path to industrialisation
6
2
18
fully migrated to workflows - our use case are AI agents that execute over a multiple-day time frame, making multiple outbound voice calls and processing the results.
before:
- scheduling service
- queues
- workers
- cron jobs
now:
- 5 functions in one file
🤯
2
1
13
the thing to pay attention to in web dev right now are workflows. it's very hard to write complex software with external service dependencies without relying on queues and workers, until now. with workflows it's 10x easier and all underlying infra is abstracted away. the paradigm has existed for a while, but it's now b...
1
3
flew 1500 miles so my friend and I could do our YC application video in person, then I’m flying back tomorrow
it would have taken a roman legion 40 days to walk this distance
it would have taken a stone age human > 3 months
the kinds of things that cross the ROI threshold in the modern world are mind blowing
4
21
1. great play by the Farcaster + Clanker teams
2. must suck to be Rainbow right now lol
We’re excited to share that Clanker is joining Farcaster!
Clanker is the leading launchpad on Base and home to projects like Bankr, Noice, Bracky and more. We’re thrilled to join forces to make it easier for everyone to create and participate in onchain communities. We plan to integrate Clanker more deeply into the a
2
13
things openai atlas browser can do with zero handholding:
- find the most optimum flight to take and hotel to book, write an itinerary doc, book the flights (with confirmation) and block out calendar
- research nearby shops with a given item in stock (on their website) and find the nearest that is still open
- QA your...
4
I was talking to a Mandarin Chinese language tutor who has taught hundreds of students, and he told me that the biggest predictor for a foreign student to reach fluent level is:
- having zero expectations
- working mostly with light effort
- focusing on building a rhythm
- optimising for enjoying the experience in the moment
7
5
27
things creators have to grapple with in a world of creator coins:
- financial markets: their success is tied to stock markets, since ethereum is tied to stock markets
- fragile fanbase: if (when) the coin crashes for whatever reason, fans lose money and suddenly morph into haters
- backseat drivers & “wen this”-ers...
3
2
13
i’m surprised that the genius, high achieving founders that I meet typically have never touched a single philosophy book. the closest proxy is usually the highly processed self-help/business books.
i have this feeling that there’s a pretty high chance that any of them would have their mind blown by many of the classi...
8
1
22
“do things that don’t scale” used to mean “don’t start by writing software”, and that was good advice when software was hard to write - much easier to validate ideas manually, and then automate yourself away
as human effort in writing software trends to zero, writing heaps of throwaway software to test ideas suddenly ...
2
1
15
in the last few months I find myself trying some consumer apps, and when I find that they don’t do exactly what I want - I think to myself: “maybe it’s worth spending a weekend to vibe code the perfect solution”
once “a weekend” becomes “a few hours”, all low stakes everyday software will be fully commoditised
I alr...
1
10
things that give my life meaning right now:
- give as much love as I can each day to my girlfriend
- invest deeply into all friendships/platonic relationships - work, everyday interactions, etc
- express myself creatively - currently build software, but also paint, make music, etc
- mindfulness/attention to all exp...
2
18
one mistake i keep seeing product builders make is conflating their big vision with the path to get there - or the here and now.
today i spoke with a founder who pitched me the ultimate "big if true" idea (a large, network-effect-driven app store kind of thing). then he simply told me i should hop on the beta and star...
3
5
27
an attempt to explain meditation as plainly as possible:
remember the last time you were so engrossed in watching a tv show, or movie, that you forgot you were looking at a screen and actually believed yourself to be in the world of that show or movie? then, at a certain moment, you snap back into the more "correct" r...
1
1
7
what you do must be fun
when I work, I work for its own sake. I’m picking the kind of work to do such that I enjoy the doing of it, itself. that way, when there are endless things to do - I’m like: “hell yeah!”.
when I get good at something, the only guaranteed reward is having more of it to do. if I instead were to...
3
4
25
objective/intersubjective accomplishments like the size of the rectangle you can afford to sleep in, how smooth your cat is, how big your sky chair is and whether your get your sky meal in glass or plastic aren’t hugely correlated with the quality of the subjective experience of your life
yet for some reason we like ...
4
10
we are about to start working with a multi-billion-$ company to reimagine their operations with AI. it’s wild how we got here so fast, but here we are
we’re a team of 2 - both cracked, both previously exited, keeping the team super small
we are based in europe, well funded with years of runway. we want to build a su...
6
1
26
this feels like a weird thing to do: pit token holders against the project founder. if the clanker team doesn’t want this, it will 100% be seen and spread as fud by a vocal minority
I don’t know the full story but atm this is making me want to uninstall rainbow
3
10
"winning" conflicts loses all meaning and value when you realise that most conflicts are the result of a psychological flaw that makes us subconsciously mimic an amplified reaction to any inbound attack, and this is so deeply ingrained in us that we often just do it on autopilot.
if someone's dunking on you, they're ...
3
2
18
the media system pays people to describe the world in the most stress-inducing way possible, and there are tens of thousands of people who make it their job to stress people out and earn huge wealth from doing so well
this nurtures a whole generation of people who adopt being stressed and disappointed as their identit...
2
1
6
they're gonna ctrl-z back to the original twitter algo
Nikita Bier
@nikitabier
7mo
The goal for your X timeline is to get out of the mainstream algo and the political crusades and find your niche.
You should be able to post about your interests and have friendly, relevant people chime in.
If you’re seeing gas station fight videos, your account is not ramped
3
10
GummySearch is a wild way to do market research. It's a massive fresh reddit dataset with an AI side-car for tracking trends, audience pain points and feature requests. We've been using it a lot in the past week to find startup ideas in the supply chain space.
Someone will definitely build something like this for Far...
3
2
some ideas I found interesting:
- thinking of consumer (and even b2b) as a competition to capture "verbs" - like watch (youtube), listen (spotify), host (aws), date (tinder), etc, etc. AI isn't going to change the existing verbs, but it might create a whole set of new verbs that are addressable with technology.
- AI ...
I think it's because actually working hard on long term things isn't rewarded by the market. It's a much weaker signal here compared to anywhere else. In crypto, you get rewarded if you're performative about what you do (hence hollow partnerships, shilling, "narrative", "roadmaps"). Current incentives pay you to larp.
1
8
honestly, so much easier these days to build your own internal tools rather than configuring something off the shelf.
example: AI-native agent eval runner/test harness. there are dozens of off-the-shelf products but I was able to write one in ~30 minutes that does exactly what I want and that I can easily extend in wh...
2
1
7
there's so much depth to reading that I missed in the first quarter of my life, where I just:
- read for some amount of time, linearly in the current book
terrible comprehension, forgot nearly everything. now, I:
- read slowly, take margin notes
- put down the material frequently to do secondary research and write m...
1
9
looking to interview people who fix stuff
1
4
going to south america for the first time in ~100 days. challenging myself to pick up ~A2/B1 level Spanish from scratch. did this with German over the last 2 years, and Spanish is likely to be much easier (straightforward grammar, lots of similar words) so I should get pretty far in 3 months.
strategy for the first 2...
3
10
I learned so much from repeatedly questioning who I am. For most of my life, the answers were all adjectives in an internal language description of my "self". In the last year or so, I started breaking down the use of language itself in my own self-descriptions. Many things previously solid started falling apart.
1
11
we picked up a hitchhiker in the middle of the desert. Austin spent the last year walking coast to coast from Florida to California. the only thing he had on him was a bottle filled up with murky canal water (the only water source for dozens of miles). The guy used to be a flight attendant for American Airlines based o...
1
10
I come back to this app because of interesting conversations + most people here don’t just want to dunk on each other.
Insta: no interesting conversations + no dunking
Reddit/HN: interesting conversations + dunking keyboard warriors
X: no interesting conversations + dunking olympics (fun to watch tho 🍿)
5
21
I’m anxiously motivated to spend a lot of my time reading and thinking because, at 31, my neural plasticity likely tapered and will gradually start deteriorating - meaning my brain will begin crystallising and new things will be harder and harder to learn.
I spent last night teaching a couple in their early 60s how t...
8
1
32
you can tell what someone really cares about based on how much time they spend talking about it and the specific language they use on social media
the structure of how they talk eludes to a certain fingerprint, dialect of communication that communicates the subconscious, and this can be very telling. often times there...
6
