Most available Personal Finance tools suck.
Their “one-size-fits-all” approach means rigid dashboards, random charts, and rote finance advice based on general assumptions - NOT advice that builds around YOU. The end result? Advice that looks right on paper but feels wrong in
It’s impressive but not as impressive as you think. I maintained a private set of math competition level (AMC 10) questions and I have been testing the model on this set. It’s performing much worse compared to the public ones.
🚀 Big news: @Invisible_HQ is joining forces with @perplexity_ai !
We’re combining our expertise in multi-agent orchestration with Comet, the agentic browser, to shape the future of infrastructure for AI agents.
This is just the beginning.
getinvisible.com/articles/invis…
Upgraded both of my car to FSD 12.5.4 (1 from 12.3.6 and 1 from 12.5.1.2). It has been terrible so far, lots of phantom breaking and indecisiveness. @wholemars do you see the same issue?
Every AI lab has their own web automation agents, from OpenAI's Operator to Anthropic's Computer Use and a million open-source clones. But would you trust them with your credit cards?
x.com/karpathy/statu…
A couple weeks ago, Andrej Karpathy tweeted about a theoretical input-conditional, on-demand GUI, like something out of Iron Man or Minority Report.
Our team has been heads down on that very idea for months. With Taka, the vision is no longer science
"Chatting" with LLM feels like using an 80s computer terminal. The GUI hasn't been invented, yet but imo some properties of it can start to be predicted.
1 it will be visual (like GUIs of the past) because vision (pictures, charts, animations, not so much reading) is the 10-lane
Let me nerd out for a second & go under the hood. What excites me most about Taka is that it’s one of the first true examples of “just-in-time” software.
Everyone’s talking about vibe-coding where AI agents help developers build software 10x faster…but the software still has to
Here's the link to the paper: getinvisible.com/articles/human…
It explains how our automation agent is built and why "agent cardinality" could be a new scaling dimension in AI.
In our new paper, "Human-Inspired Agent Design in Web Automation: From Principles to Practice," we show how our multi-agent approach surpasses OpenAI Operator and Claude Computer Use by a wide margin and hit a state-of-the-art 88% on Mind2Web.