Scriber
submittedBuilt Scriber, a two-agent voice architecture for live standup facilitation. One agent runs the meeting in voice and mutates the Linear board live; a second silent agent remembers across standups and whispers guidance.
Built in one night for OpenAI Voice Hack Night (Cerebral Valley, May 27 2026): a two-agent voice architecture for real-time standup facilitation.
Every existing meeting AI (Spinach, Otter, Fireflies, Fellow) is passive-then-active: it listens, then acts after the meeting. Scriber flips it. The agent acts during the conversation, with the Linear board visibly mutating live. The visceral demo moment: someone describes a flow, Scriber generates a diagram with gpt-image-1, and the image appears on the Linear ticket before the speaker finishes their sentence.
- Scriber facilitates in the
cedarvoice (gpt-realtime-2, low latency): joins the call, prompts each speaker, executes tools live. - Mnemo never speaks into the call. It runs in parallel, holds long-term cross-standup memory, and whispers tactical guidance to Scriber over a side-channel.
Stack: OpenAI Realtime API, Linear GraphQL (ticket mutations + 3-step file attachment), gpt-image-1 for live diagrams, Slack webhook for the recap. Live at usescriber.com.