Commercial in-call AI assistant designed to join Zoom and Google Meet conversations both passively (listen, record, transcribe, summarise) and actively (speaking AI agents with role personas — architect, business advisor, legal consul). Per-participant language detection, host-only live subtitles with translation overlay, and live in-call insights surfaced from a private knowledge base. Cross-meeting memory of participants. The name follows the Latin tradition (adsessor — an assistant judge or expert advisor seated beside the magistrate during proceedings), in line with other ecosystem products (Verdicus, Scrutator, Consilium, Munera, Transcribator). Currently in Phase 0 — design validation. No production deployment yet.
Capabilities
- Passive baseline: listen, record, transcribe, summarise (Zoom + Google Meet planned for Phase 1)
- Speaking AI agents with role personas — architect, business advisor, legal consul
- Per-participant language detection + multi-language replies
- Host-only live subtitles with translation overlay (independent of what other participants see)
- Live in-call insights from a private knowledge base (RAG over past meetings + ecosystem corpus)
- Cross-meeting memory of participants (faces / voices / preferred language / past topics)
- Ecosystem-native: identity via Auth Arcana, LLM routing via Model Connector, retrieval via Scrutator
- Direct platform integration — Zoom Meeting SDK + headless Chromium for Google Meet (no third-party bot vendor)
Current autonomy level
Weakest link
Phase 0 — design validation only. The 5-day Zoom Meeting SDK spike is in progress to confirm raw-audio capture (PCM 16LE) and cross-account bot flow on the target infrastructure. No customer-facing code, no marketing-site content yet — adsessor.app is a placeholder until Phase 1.
Roadmap to L4
- Phase 0 → Phase 1 — finish Zoom SDK validation + ship passive baseline (record + transcribe + summarise) for Zoom under a GDPR-aligned bot disclosure mechanism.
- Phase 1 → Phase 2 — add Google Meet bridge (headless Chromium) + live subtitles with translation overlay.
- Phase 2 → Phase 3 — introduce speaking AI agents with role personas backed by the ecosystem knowledge engine.