Personal agents' best use cases are low-frequency; Muse retention drops
The author argues personal agents' top use cases (finding savings, recovering funds, rate shopping) are high-value but low-frequency, cites Sensor Tower showing Muse retention falling from 73% day 1 to 22% day 14, and complains it sends about five suggestions a day.
pattern across personal agents + fintech: #1 use case seems to be finding + executing savings (agentic rocket money), retrieving lost funds, + point-in-time rate shopping problem is these are high value but low frequency. can only cancel a subscription once + once ur insurance/card renewals are shopped, you're not gonna think ab them for another 12 mos prolly part of why Muse retention is falling off (73% day 1 to 22% by day 14 per Sensor Tower, now below ChatGPT) + also part of why Muse is low key harassing me (sends like 5 messages every day with ideas for things it can do for me) - proactive doesn’t help if the actions suggested aren't worth doing 🤷♂️
This post quotes another post: x.com/jbahrdestefano/status/2110429774933901326