The visual methods tend to feel extremely complicated – despite the under lying logic being simple. We’re working on a third approach: using LLMs to transform the representations. Love to chat and exchange notes here!
@jiquanngiam
-
Chat UI limitations and the case for direct action buttons
By
–
Chat as a primary user interface has its problems: It's "action at a distance" — the messages don't come with context on where in the UI I'm paying attention to. Getting the prompt right shifts the burden to the user, not great. Sometimes, I just want a button to press –
-
Benchmarking Lutra_AI: Request for Technical Resources
By
–
Nice! I’d love to benchmark this against @Lutra_AI too— do you have a pointer to run the benchmark?
-
Empowering creativity through AI building blocks and tools
By
–
Love the feeling of giving people lots of building blocks and empowering their creativity
-
Agent AI Interpretability: The WYSIWYG Problem Explained
By
–
Agents have a WYSIWYG problem. For text, image, and video generation — the generated artifact is something you can look at and at a glance determine quality. For agents, the series of actions they take and the intermediate results is what they produce – often opaque.
-
Transparency in AI: Making Plans and Actions Understandable
By
–
WYSIWYG = What you see is what you get. The more you can make the plan and series of actions transparent and understandable — the more users will be able to succeed at using them.
-
Email AI Tool Needs Markdown to HTML Formatting
By
–
Love the seamless nature of using email! I did find the response came back in badly formatted markdown though – may be worth having a markdown to HTML renderer for it. Great concept!
-
Linear Decoding Breakthrough Achieves 1M Context Length
By
–
This looks super promising — linear decoding complexity while improving on longer context lengths up to 1M! Super interested to see how this does when scaled up.