perpetually switching context https://xitter.cf/t.co/VOeutnlzFU (WIP) https://xitter.cf/t.co/faRyN5GPBc ($106)

Joined December 2024
I’ve been enjoying the time I spend on X lately. There’s a small crowd of people I repeatedly keep running into, I read their posts, they respond to mine, and we end up talking about tech, AI & whatever else catches our attention. Then there are all the people that I believe have made it, I see them build aspirational products, and come up with thoughts that I wish I had first. I really like being around that. But all of these interactions are currently limited to a few replies. I’d love to see what could come out of one-on-one conversations rather than ones where our attention is distributed across 200 things on the timeline. Maybe a friendship, something we build together, or a conversation that sparks a deeper thought. So I’m opening up two slots in my calendar every day to talk. I'm not really sure if I will have any takers, but feels worth it to put out there. For all of you, who don't know me here is the TLDR. I’m based in Bangalore, India. I have 3 degrees in physical science, sociology and business all from the best colleges in India. I worked in business roles at early-stage startups before starting an automation agency two years ago. We help small businesses automate the gruelling parts of their everyday work. We’ve grown mostly through referrals and partnerships, with $0 spent on marketing. I can write code, even without LLMs (though I don't anymore 💁). A big part of my work is taking messy descriptions of business problems and turning them into tech solutions. I'm also deeply interested in philosphy and my only consistent hobby is reading. The agency the food on the table, but I don't think I want to run a service company forever. I’m determined to build a product company. I just haven’t figured out what. With the constant idea search, I'm now in a little bit of a state of paralysis. I work remotely, so a lot of this happens alone at home, in front of a laptop, where I'm talking to agents most of the day. I’d like to make room for conversations with people whose worlds look different from mine or even exactly the same. We can talk about anything, an idea, a problem, something you’re building, or just something you’d enjoy talking about. Or you can also just come in with a blank slate, I yap a lot, so I'm sure you'll have something to talk about once you are on the call. Two slots a day. Book one here: calendar.app.google/zFKF6sYw…
7
2
22
1,602
While AI has been solving math problems we don't even understand, some kid from UChicago solved dating. He has shown mathematically we can't all date out of our league
4
73
Today, for the first time I trained a client's back-office team of 20 people on how to use Claude. Basic stuff like scheduled tasks, skills, browser user and co-work. This has been the most rewarding experience I've had. The sparkle in their eyes was totally worth it. I feel like I'm a magician!!
3
8
159
The intellectual decline of our entire species is not really far fetched anymore. There will remain a few who will still be using their minds, but the rest of the hive will probably not even know that they are being controlled by the ones with mental capacity intact.
We're about to enter a period where not reading or writing becomes as ubiquitous as smoking once was. We can hope people will eventually realize this is bad for them, as they did with smoking. But this is far from certain; the cost may not be as obvious.
1
33
The best sentence I’ve ever read in English- “I was too busy listening to hear”
2
42
My timeline is full of releases, launches and funding announcements, day in and day out. This is one, is probably the most important ones I’ve seen recently. Go read their original missions statement: distributed.inc/research/our… Here are three reasons why this is so impressive: 1. They want people to own and control pintelligence. As we move in the direction of offloading our conginitive capablities to AI, this becomes especially important. (remember all the cyber security news going around this week) 2. In their experiments, well-organised team of smaller AIs outperformed its strongest member. Getting them to work together and communicate well was key to this. Hence they are building the communication layer. 3. Diversity matters. Human dominance over nature did not result from one super-smart person, but rather from our diverse collective intelligence and action Still early, and plenty to prove. But I'm really excited for the direction they are taking.
Today distributed cognition exits out of stealth with backing by Verdict Capital. Our mission is to replicate humanity's collective intelligence. Read more: dcog.com/research/our-impera… Most of today's AI power is held in the hands of a few, who can choose what you're allowed to think and what you're permitted to do. We stand on three substantially different foundational beliefs: - You shouldn't ask permission for intelligence. - Superintelligence will be a distributed collective. - Humans are not meant to be relegated to a permanent underclass. If we don't distribute intelligence, nobody will.
1
60
Deepseek go brrrrrr
DeepSeek V4.1 Flash 1017 runs at 1,521+ tok/s on Fireworks Ultrafast. Fast enough to ship Tetris in 11 seconds, 7x faster than Claude Opus 5.5. Join the Ultrafast waitlist: bit.ly/3UA0wKq
2
3
70
Dot managing all my work across projects on Notion is the best EA ever. I can't believed I once hired an EA of Upwork to do the same work. Insane times!
2
129
I understand when enterprises had adoption problems with legacy software, like Netsuite ERP, Jira, and Salesforce. They were clunky, people had to learn a specific way of working and they had to adapt their work to it. But, how can there be adoption problems with Claude???? Open the chat window and type, or even better, speak to it in voice mode. I just had a call with the CIO of a prominent enterprise, they pay for 500+ Claude seats and less than 10% are used everyday. I don't know how to comprehend this.
4
6
300
With Anthropic's new API version, the change getting most attention is that it drops temperature. But I don't think this what affects the builders the most. I installed the new SDK and tested what this actually takes away. Nothing really. >> Opus, Sonnet and Haiku 5.5 already rejected any temperature value, except the default on the old SDK with a 400 from error. >> Sonnet 4.6 still accepted it, so I asked it for the same one-sentence summary 15 times at temperature 0 and 15 times at the default. Temperature 0 gave me 3 different wordings instead of 5. Fewer, but not repeatable. A one-word label came back identical all 30 times. The upgrade is where things broke. I installed pydantic-ai fresh, it pulled anthropic 2.1, and my first call died with "TypeError: Cannot instantiate typing.Union" before it reached Claude. Pinning anthropic<2 fixed it. If your framework doesn't cap the SDK version, make sure to pin it yourself.
API version 2026-10-21 is now available. The Claude API has evolved a lot, but this is our first official version release since 2023. We plan to release versions more often so you can handle breaking changes before a new model requires them. 2023-06-01 is not deprecated.
1
3
114
Tinyfish is one of the best search tools out there. Have deployed it in multiple projects and it almost never crosses the free tier limits. (vv generous free tier) Highly recommend!
TinyFish is now a Verified plugin on Hermes Agent. @NousResearch just introduced a Verified tier to Hermes plugins. We’re one of the first three to carry it, alongside @nvidia and @trycua. hermes-agent.nousresearch.co… Plug it in and Hermes gets free TinyFish Search and Fetch, and Web Agent when a task needs to be completed on a real website, like filling a form or getting past a login. hermes plugins install tinyfish Thank you @Teknium and the Nous team.
2
4
207
One roller coaster of a day for Postiz
1
53
Most of the people affected by today's PHP bug don't even know what PHP is. Many of my clients reached out asking if they're affected. So I made them this fun explainer. It goes out with a short report on what I've patched and what stays offline until it's upgraded.
2
1
4
222
Wow!! Absolute shame A team of vibe coders would have done a better job than the Indian Certificate authority
Gross incompetency from Indian certificate authority eMudhra. You had one job: generate a 384 bit private keys and instead you used: 4, 56, 66, 115, 155, 156, 230, and 237. This allows attackers to issue certificates for any websites. Embarrassing that it’s a family business run by largely computer illiterate people. Only 1 of the 3 have a remotely computer-related background and even they never actually worked a technical job in their life. If you look at the LinkedIn, their tech team is young, from tier 3 education and paid sort of bottom barrel. Imagine how many attacks you can design with AI given the existence of this genre of gross negligence.
1
12
524
A statement issued by CERT (Computer Emergency Response Team) warns of a compromise involving eMudhra’s eight intermediate CA certificates due to weak cryptographic keys, which could allow attackers to issue fraudulent TLS certificates and intercept or modify encrypted communications. The advisory outlines mitigation measures, including security updates, certificate replacement where required, and verification of certificate trust chains.
21
It's ridiculous how many big tech companies have great blogs, but no way for me to sign up for email notifications for when they publish new blogs.
1
5
194
Banger read!! I’m disappointed I just found this, and the archive of all past writing is too large to consume. Highly recommend!
I just published my weekly reflections: docs.google.com/document/d/1… Software pyramids. AI-made sandcastles. Dead-end software. Spending tokens on opex vs capex. The Reset Nexus. Mentorship trust fall. The AI Peter Principle. The non-viable solution to Goodhart's law. AI making your spikes sharper. The cost of linearizing writing. The Lethal Trifecta Gordian Knot. Chats as a junk drawer.
1
2
91
I hate going to sleep nowadays. I had 700 more ideas to try. Claude is already running 200 sessions over night, whilst I sleep. How does one make time to try out all ideas one has. And also make time to go to the gym, and talk to people and have a girlfriend. (Numbers above are more reflective of how I feel and not actual counts)
2
3
90
You Can Ship Anything
You Can Ship Everything
1
49