IHSAN · PRIVATE VOICE PILOT
Talk with Mehdi.
A natural conversation with your AI assistant. Choose a voice, then try a conversation. You can interrupt at any time.
Haytham is the current voice; you can change it any time.
Compare the three English auditions in your ElevenLabs Text to Speech history. They use the same script and model; replaying them does not generate another sample.
Checking readiness…
Choose Mehdi’s voice
Three male voices: Hamid, Anas and Haytham, each generated speaking English. The Chrome group “🎙️ Mehdi English auditions” also holds their players. Use the History tab to replay each named voice; Haytham is selected for now; Hamid and Anas are restricted to recorded auditions by ElevenLabs. The English audio files have not yet been exported to this page.
Conversation timing
Event timing appears here after a real call. The measurement is transcript receipt to the agent-speaking event. It is not speech-end latency or an acoustic benchmark.
No live measurements yet.
Knowledge and deeper work
During a voice call, ask Mehdi to retrieve evidence or start deeper analysis. Source citations and job updates appear here. The deployed Ihsan backend uses public sources and a live Codex worker; Claude availability is reported separately. The owner token also enables typed testing.
No query yet.
Voice tools use scoped, short-lived authorization. Longer jobs continue independently of the call. Native Slack Huddle participation is still pending. Voice sessions end after five minutes.