@dannypostma Most of my agents run locally because my laptop runs close to 24/7. I specifically run some tasks remotely. And some tasks are offloaded automatically to remote server.
@elliotarledge i tried this and it's 4x faster than Whisper large-v3-turbo! it does very well on plain english but i'm seeing way too many mistakes on dictionary words compared to Whisper large-v3-turbo.
@elliotarledge i have hours of recording to test this on. Whisper large-v3-turbo is the best fast local model i've tried. i'm testing this now and i'll let you know if it beats Whisper.
Background subagents in Claude Code now use less of your usage limit.
A running subagent updates its status line every 30 seconds, and each update used to re-read its whole conversation. Now Haiku 5.5 writes it from short notes instead, so each update costs about 99% less.
@thorstenball does the same agent deploying the feature watch it rollout?
does every person of the team have this setup? or do you also have agents watching every deploy across the team?
@thorstenball screen recording to agents is great! do you copy paste the recording to a specific agent? or do you have an agent deciding where the recording goes?
@thorstenball yes. every token i spend on agents verifying their work end to end is the best spent token.
tests can be useful because they tell the agent about behavior they are changing. but i would rather delete all tests than take away my agent's browser.