Nice product.
I pulled the README and I think the main opportunity is just hierarchy.
Right now the first screen is carrying the whole surface area at once: dashboard, CLI, replay, conversion, handoff, five agents.
All of that is real, but the sharper story feels simpler:
this is a control room for AI coding sessions.
That job lands fast because the pain is already familiar:
people are juggling sessions across tools, can’t find the right thread quickly, and end up reconstructing context from logs.
I’d consider leading with something closer to:
“Control room for AI coding sessions. Search, replay, and resume Claude Code, Codex, Cursor, OpenCode, and Kiro sessions without digging through scattered logs.”
Then use the multi-agent matrix as proof of breadth, instead of making breadth do the opening job by itself.
The rest of the README already has the proof. I just think the first screen could land the core job-to-be-done faster.
If useful, I can mock up a copy-only top-of-README patch.
Nice product.
I pulled the README and I think the main opportunity is just hierarchy.
Right now the first screen is carrying the whole surface area at once: dashboard, CLI, replay, conversion, handoff, five agents.
All of that is real, but the sharper story feels simpler:
this is a control room for AI coding sessions.
That job lands fast because the pain is already familiar:
people are juggling sessions across tools, can’t find the right thread quickly, and end up reconstructing context from logs.
I’d consider leading with something closer to:
“Control room for AI coding sessions. Search, replay, and resume Claude Code, Codex, Cursor, OpenCode, and Kiro sessions without digging through scattered logs.”
Then use the multi-agent matrix as proof of breadth, instead of making breadth do the opening job by itself.
The rest of the README already has the proof. I just think the first screen could land the core job-to-be-done faster.
If useful, I can mock up a copy-only top-of-README patch.