Agent-in-Call

Definition

An AI agent that joins a customer call in real time to answer product questions the human seller / CSM doesn’t know — or to suggest next-best actions. Proposed by saar-arbel in the Atera interview and validated by multiple external customers.

Three external validations to date:

1. rachel-rembrandt (Atera, 2026-04-19) — “an amazing idea”; her CSMs already ask Claude product questions mid-call.
2. anthony-landsberg (Cerebras, 2026-04-20) — contrasting data point: not yet using voice agents; prefers traditional phone. Validates that high-volume AEs are not the early adopter.
3. shirin-aalami (AAICO, 2026-04-23) — already does this manually: queries Gemini mid-call for product FAQ, security specs. Prefers text over voice for real-time AI. Strongest validation of the core behavior.

Related decisions

(None yet — not committed to roadmap.)

Sources

• atera-2026-04-19 — first external validation (Rachel)
• ai-for-gtm-cerebras-2026-04-20 — contrasting data point (Anthony; not using voice agents)
• ai-for-gtm-shirin-2026-04-23 — third validation; Shirin already does this with Gemini (text preferred)

Related entities

• saar-arbel — proposed the concept
• rachel-rembrandt — first external validator
• shirin-aalami — third validator; already lives this behavior with Gemini
• anthony-landsberg — contrasting: experienced AE not yet adopting voice agents

Related concepts

• brain-layer — the agent reads from / writes to the brain
• self-learning-voice-agent — adjacent product surface; could share the voice stack
• cross-funnel-intelligence — the agent can pull cross-funnel context into a live call

Open considerations

• Use case extends beyond SDR — applies to CSMs, AMs, sellers
• Trust / accuracy of answers (Rachel flagged: pricing, exact feature capabilities are tricky)
• Text interface may be preferred over voice for real-time assist (Shirin’s preference)
• Integration: Notetaker + brain layer + product knowledge corpus