what if a webpage had vim like motions? well now it does. Found a chrome extension called Vimium that adds two to three character tags to every clickable surface on a page. You enable it by pressing f
https://t.co/H7m5tzgNAm
I like keyboard shortcuts. One I had to build was to switch through visible windows of a particular app. Now I can press Ctrl + 9 and cycle through all my Brave windows :)
https://t.co/BUCVd0nPUd
Read up on WebMCP a little, I think it will change how frontend can be built to be agent first.
Core idea: if a website needs to be used by agents, the only method is via raw JS based actions. The webpage doesn't expose a set of tools for standard expected actions that are supported by the UI. The webpage didn't even have a mechanism to do so even if it wanted to.
WebMCP adds the ability to register tools with an MCP surface for your agent to interact with the website. It is different from a MCP servers where tools execute on the server, as WebMCP tool execution is entirely in the browser.
Right now it is an experimental flag in chrome:
chrome://flags/#enable-webmcp-testing
Worth reading: https://t.co/4Jisb557Sl
What if software in the future just has codex always available as a thread that can be notified on errors and the code be live patched to deal with the issue XD
The era of bad software will give birth to the era of self healing software
Pointer is hiring!
We're growing faster than we can handle and need people across every function.
We build AI systems that do a company's hardest work and prove it was done right, deployed in some of the highest-stakes operations in the world.
If this sounds interesting, learn more at https://t.co/xyf2mkkHdL. Refer someone we hire and we'll pay you $15,000.
We at @pointer hosted a dinner with @stepstonegroup at Le Pavillon in NYC last week for a group of finance leaders in banking, insurance, fund admin, and more.
It's clear that nobody wants AI they have to babysit. The systems delivering the most value are the ones that can remove the human reviewer entirely and run in the background for hours.
More soon!
@Henhound@nealchopra to be cost efficient we decided to stay in the anthropic family - gpt 5.5 / opus 4.7 have relatively close performance. staying in same family allows us to explore the nuances of the models more closely
Today, we’re sharing a new state of the art for computer use.
Our system holds the two highest verified scores on OSWorld, the standard benchmark for AI agents that operate a computer like a person: 83.6% using Claude Opus 4.7 and 81.5% using Claude Sonnet 4.6. The human baseline is 72.4%.
🧵 1/7