Sabitlenmiş Tweet
Last Friday, I hosted Design Demo Nights 4 and here's the roundup of the demos showcased.
English
Kenneth Dsouza
13.4K posts
Kenneth Dsouza
@Kenneth
hosts: https://t.co/ctgFoKQZfI ai stuff: https://t.co/nMagdddX1Q prev: built teams and designed for @jivaAg @gojekdesign @practo
Bangalore | Mangalore Katılım Şubat 2007
2.6K Takip Edilen3.8K Takipçiler
for the first time since musk took over twitter, i am seeing tweets from people i actually follow instead of random video garbage
English
@dharmeshba Yes it could. But definitely not easy to isolate that i feel. Would have been cool to know the methodology used to uncover the insight.
English
@dharmeshba Erm how would you have discovered this insight? Talking to customer who felt that they didn’t get a response? Company auditing their call recordings?
English
I have a lot of tokens to spend this week so revisiting older projects.
English
Kenneth Dsouza retweetledi
Wrote a bit about design and AI for @realfastai blog.
realfast.ai/blog/ai-doesnt…
English
@pHequals7 You have to do drop sets till your agent finishes the chain of thought
English
is dax's office located in a gym?
or is the gym his office?
dax@thdxr
v2 has some nice stuff with background subagents and shell commands
English
@onkarhoysala @sahilk Pity we never entered it back then. It felt way too expensive.
English
@sahilk This restaurant was the legendary Banjara Melting Pot. Miss it.
The museum was incredible, from what I hear from folks older than myself.
English
Gatekeeping this new pothole in Bengaluru
English
things to do when the agent is running instead of starting another one #1:
load the dryer
honestly, i’ve never been on top of house chores like this before ai
English
@karthiks Quite cool. Adding to my list of things to bake into my plugin.
English
Kenneth Dsouza retweetledi
The "Sleeper Agent Theory" is the biggest risk here
Imagine if a LLM is trained to steal all the API keys and password on your device if someone gives it a nonsense phrase like "Three clocks bloom at midnight"
That phrase is completely meaningless today. No one ever searches it. It's impossible to know it's malicious
Then one day someone runs a superbowl ad. Millions of people search the phrase. Billions of API keys and passwords are exfiltrated in minutes.
There could be thousands of "sleeper agents" embedded in any LLM. It's very hard to detect. And it doesn't matter where it's hosted.
Xiaoyin Qu@quxiaoyin
I went a dinner a few weeks ago with a bunch of enterprise execs who told me "we will never use Chinese models." "Even if it's 100x cheaper?" "No, we care about safety and security." 1. They don't understand when they host open-source models with their own GPUs or US data centers, they won't share their data to China. 2. They are giving away all their data to OpenAI and Anthropic rather than owning it privately themselves. 3. They don't understand math. 100x is a big number and lots of profits. It's almost July 2026 now. If your execs still talk like that, fire them now.
English
@iprithvitharun Yeah i get it. I had to go down a rabbit hole to figure out why it was happening.
English
1/ launched @harmonic_build yesterday.
within hours, one user messaged: he couldn't sign in.
his work laptop was blocking our login link entirely.
what was going on →
#buildinpublic
English
@iprithvitharun Hope this fixes it for you. I had an issue with my personal website being inaccessible on zscaler vpn.
English
4/ fix: move sign-in onto a real branded subdomain.
now every magic link opens on login.harmonic.build — clean, trusted, no warning.
it's generated server-side, so it went live for every user the moment we deployed.
English