@tankots@WisprFlow Editing to add to dictionary only works ~30% of the time ๐
Also finding I sometimes switch back to Apple dictation now, ONLY bec I can see live progress as I talk
@dunkhippo33@vishakha041@computextaipei Huashan 1914 has tons of cultural events and is same metro line.
ๆฑ้ Dongmen station area has tons of food & cafes.
ไธญๅฑฑ Zhongshan and ่ฅฟ้ Ximen are where all the young ppl flock to
Xiangshan ่ฑกๅฑฑ has a good easy hike next to Taipei 101
@Alfred_Lin Itโs tough if youโre on a team where most ppl hold prior default. Theyโd even ridicule you. Really important to hold firm and keep re-examining
@thisisgrantlee So good โค๏ธ. I come back to joy. Working in @GammaApp brings me joy, and I know tons of users who feel the same. That "feeling" part is often underrated.
@interaction The onboarding was ridiculously easy and smooth. Only (tiny) glitch was Google integration but quickly resolved. This is what it should feel like to begin using an agent product
@karpathy@peterxing Yeah even when I share code, my agent opts to distill the idea anyway and finds what's applicable to my own knowledge base (did it w your first tweet)
And itโs insanely hard to set up. I had @claudeai code do it. 1 hr, 5 debugging sessions later, still not working. Claude code canโt find why. 1st time Iโve felt an impossible barrier w cc
Anthropic's Claude Code Channels supports three messaging apps. Telegram, Discord, and iMessage. That's the full list. The community has "requested" Slack and WhatsApp.
OpenClaw already runs on 20+. WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Microsoft Teams, Google Chat, LINE, Matrix, Feishu, Nostr, IRC. Tencent built a full WeChat integration on March 10. Nearly 1,000 engineers lined up at their Shenzhen headquarters to install it on their laptops.
By shipping OpenClaw as an MCP server, Steinberger just made that entire messaging layer pluggable into any MCP-compatible agent. Including Claude Code. A developer can now route Claude's brain through OpenClaw's 20-channel infrastructure and skip Anthropic's three-app limitation entirely.
This is the move Anthropic designed and can't block. They invented MCP. They donated it to the Linux Foundation in December. They built it as an open standard specifically so third parties would build connectors. Steinberger is now the third party building the connector that makes their own Channels product redundant.
The sequence: Anthropic forces a trademark rename on Clawdbot. Steinberger joins OpenAI. Anthropic ships Channels as the "OpenClaw killer." One week later, the guy collecting an OpenAI paycheck turns Anthropic's own protocol into the reason developers don't need Anthropic's product.
Three companies. One protocol. Zero way to stop it.
The new Gamma Imagine is insanely good at graphic generation ๐คฏ
I gave it a date and a theme for an event and asked for an invite.
It came up with three genuinely clever options - and I could edit each with text prompts + easy buttons ๐
@julianlehr I agree words are slower. But I like chat for 2 reasons: it allows me to focus, bec GUIs are often designed to keep me there. Eg Linkedin
And words are better for nuance or brainstorming, like you wrote in the post
Truly wild story ๐คฏ. A new era of "citizen science" is beginning.
An engineer with no medical training used ChatGPT and Googleโs Alphafold (AI protein sequencer) to build a working cancer vaccine from scratch.
He turned raw genetic data into a custom mRNA vaccine that shrank his dying dog's tumor by 50%.
Paul Conyngham spent $3000 to get the DNA sequences of his dog's healthy blood and the cancerous tumor.
He was staring at gigabytes of raw genetic code without having any clue how to read biological data.
This is exactly where ChatGPT became the crucial missing link in his process.
He used ChatGPT as a high-level biological consultant to figure out how to compare the two DNA samples and spot the exact mutations causing the cancer.
ChatGPT gave him the step-by-step instructions to run the data pipelines and pointed him toward an AI tool called AlphaFold to map the physical shape of the damaged proteins.
The chatbot basically translated complex oncology concepts so he could write a half-page chemical recipe for an mRNA vaccine.
This mRNA is just a genetic instruction manual that tells the immune system how to recognize and attack those specific mutated cancer cells.
University researchers were blown away by his formula and manufactured the physical vaccine for him.
A veterinary expert then injected the dog, and within weeks the massive tumor had halved in size.
Iโm surprised by how little โproprietaryโ users Claude has. Maybe a function of prosumer users mostly using everything? Eg Iโm now 90% Claude and 10% every other LLM
3. ChatGPT owns the most "proprietary" users
Across the largest LLM players, ChatGPT has by far the largest audience that exclusively using the platform.
Gemini has a base of its own - but users of almost every product are essentially "shared" with ChatGPT
@nikunj Itโs precisely some of my senior exec and programmer (!) friends who refuse to use AI. Their argument? โ Iโm fine. I know what Iโm doing.โ
@harjtaggar@rohanpaul_ai This is what Iโve felt for a couple of years: an AI powered workday is tiring as hell. More data, more questions, more learning than ever
@nikunj It surprised me how many VCs donโt use the products they invest in, or even had never used them!
Indeed itโs more important than ever to be building constantly.
Honestly tapping GUI is often faster.
BUT: my actions are often followed by updating files, index, or calendar/tasks, so telling the agent to do it ensures execution on the right files, to the right spec