We’re thrilled to announce that @FathomDotAI is now part of Superhuman!
Fathom has built an incredible AI notetaker that helps teams capture, search, and act on what happens in meetings. Bringing Fathom into Superhuman adds an important source of context to our suite—connecting what people discuss and decide with email, documents, data, and the apps and websites where they work.
We’re incredibly excited about what we can build together. Welcome to Superhuman, Fathom team!
@mntruell i think about both model performance and new uis as critical dimensions on which we’d need to continue to innovate and improve to reach the “perfect” ai coding experience. any others that are front of mind?
would be quite nice to be able to configure coding agents to account for if the user is:
• a better coder than designer
• a better designer than coder
most useful in 0->1 or exploratory projects
even with the increasing quality of llm-generated code, I sometimes find it hard to trust the changes without manually reviewing each line
code review is often my bottleneck over code generation
a thought dump on how ui could help me increase trust in llm-generated code: https://t.co/kuJsc2uD8r
interesting to think about code review uis at the prompting level rather than the pr level
how do i build a ui to increase trust in each change i prompt a coding agent to make?
context is rich at this point too. have access to:
- user’s prompt to the agent
- agent’s text response
- the actual diff
the first two open up experiments around intent alignment — did the agent understand what i actually wanted?
even with the increasing quality of llm-generated code, I sometimes find it hard to trust the changes without manually reviewing each line
code review is often my bottleneck over code generation
a thought dump on how ui could help me increase trust in llm-generated code: https://t.co/kuJsc2uD8r
@rohanvarma Do you think the ADE abstraction is abstract enough for us to build for non technical users as well? I imagine the visualizations and UX required to make an ADE useful might still be different enough for technical vs non technical users
even with the increasing quality of llm-generated code, I sometimes find it hard to trust the changes without manually reviewing each line
code review is often my bottleneck over code generation
a thought dump on how ui could help me increase trust in llm-generated code: https://t.co/kuJsc2uD8r
so interesting to see how hiring for “ai engineers” has changed in the last ~1yr
my observations:
1. shift from hiring hobbyists to hiring people who have used tools and frameworks and driven results
2. the role itself is more defined, as those hiring have better ideas of what an ai engineer on the team would actually do
weekend ship - excited to launch the women who lift website!
running more events in sf, join the community here 🫶
https://t.co/FLbGDgciUm
@khushi_shelat@shreyaw_