Loopr 用 Jev 给通知自动分级

Zach Daniel @ZachSDaniel1原推发布于 2026.10.11

在 X 上打开原推

I haven't been focused on pushing http://loopr.life as a product, but I've been quietly pushing tons of features and have been driving pretty much my whole life from it. You can configure OpenAI/Anthropic models, and even Jev for the models that auto-triage inbound notifications (set priority, lowest priority gets sent to a holding area that you can review later). Things you can now do from directly within Loopr: - Take action on calendar invites - View and manage your calendar in a daily view - Write responses to emails, linear notifications, and GitHub notifications - Unsubscribe from mailing lists/marketing emails with one button (`X`) Things that I might release: - I've built a concept of "Rings" which are shared loops between either a company and a family. You can be a member of multiple rings, and for each person in each ring, you have a shared collaboration page. You can include some rings in your own view, or cycle through each ring to see what's in it. Items in your own loop can be moved out to a ring, or you can claim items out of a ring. It's a very interesting and task oriented way to collaborate with a team. If I release rings, it will likely be on some kind of "Pro" plan. - I'm considering building a "Dot"/"Grokbot" style service directly into loopr, except built on concrete workflows that you can build and thus trust more. Instead of just "rules as context" (and therefore really just suggestions), you build a workflow using either time-based triggers or event-based triggers, and chain together decision models/programatic checks, approvals (approvals just show up in your loop). And of course Loopr has an MCP, so you could collaborate with claude or eventually your Loopr agent to build a workflow. You could configure MCP servers, it would have sandboxed network access etc. This would likely need to be usage-based-pricing for the compute, and bring your own API key (like all the other AI features in Loopr) for the rest. I already manage all of my work and signal through Loopr, so why not have my agent work in the same place and on the same signal?