Daily Edition

illmethinks — September 12, 2026

Agents are getting new voices, shared workspaces and inboxes. The harder question is who owns the action behind each friendly surface: the person approving it, the runtime executing it, or a connection that only looks ready.

Hermes, OpenClaw and NanoClaw propose GPT-Live voice fronts for their existing agents

Three projects are adapting GPT-Live to listen and speak while their existing agents do the work. The proposals differ on account routing, session ownership and caller identity; none should be confused with a generally shipped, independently tested integration.

Salesforce Ben maps OpenClaw multiplayer to team governance, not a turnkey CRM integration

Tim Combridge's September 11 analysis applies OpenClaw's shared sessions and approval controls to Salesforce work. The use cases are explicitly hypothetical; the underlying OpenClaw 2.0…

Hermes removes Collective Wisdom from core and proposes a smaller consent-gated plugin

A merged removal takes the organization skill marketplace out of Hermes core. A fresh proposal brings back a narrower instruction-sharing plugin, retaining explicit consent and hash…

Paperclip proposes persistent agent chats that hand plans into ordinary project tasks

A four-part Agent Chat stack would give each person an ongoing conversation with an agent while retaining task controls and durable project handoffs. The experiment stays off by default and…

ZeroClaw proposes checking whether a plugin can load before declaring it installed

Two related proposals distinguish a discovered WASM package from a component the host can instantiate. They would add install-time refusal and explicit verification for existing plugins,…

Hermes proposes onboarding that waits for real app authorization instead of talking around it

A Desktop proposal makes connector setup a visible lifecycle: open one authorization, wait for it, and retain the connected result. The guided first-launch flow remains flag-gated, with…