# Call Transfer & Human Handoff > How RingBot transfers calls between AI agents and escalates to humans without losing context, with handoff rules and specialist routing. Source: https://ringbot.ai/docs/call-transfer-and-handoff --- Move a conversation between AI agents — or from AI to a human — without dropping the story the caller already told. ## Why handoff is a product feature Most voice-agent demos look fine until the caller needs something the first prompt cannot handle. Then the system either invents an answer or forces the customer to repeat themselves to a person. In real businesses that failure mode is expensive: lost bookings, angry callers, and staff who inherit a cold transfer. RingBot treats transfer and escalation as part of [multi-agent collaboration](https://ringbot.ai/features/multi-agent-collaboration), not as a bolt-on webhook. Agents can hand work to a specialist agent, consult another agent, or escalate to a human while keeping conversation context across the transition. ## Handoff patterns ### Agent → agent A general receptionist agent routes to a booking specialist, billing specialist, or language-specific agent without restarting the dialog. ### Specialist consultation One agent can bring in another specialist path when the question leaves its scope, then continue with a coherent answer trail. ### Agent → human When rules say a person must take over — policy, emotion, or explicit request — escalate with the transcript context the team needs. ## What “keeping context” means in practice A useful handoff is more than bridging the audio path. The next participant needs enough state to continue: who the caller is (when known), what they asked for, constraints already collected, and why the transfer happened. Without that, every escalation feels like a cold restart. - Caller intent and key entities already captured (name, time window, order ID, etc.) - Reason for transfer so the human or specialist does not re-interrogate from zero - Channel continuity for voice and chat workflows that share the same agent stack - Clear ownership after the hop — which agent or queue is now responsible ## How to design transfer rules ### Split by ownership, not by vanity prompts Give each agent a job it can finish: intake, scheduling, FAQ, sales qualification. Transfer when ownership changes — department, skill, or compliance — not when a prompt gets long. ### Define explicit escalation triggers Typical triggers include caller asks for a human, repeated failure to resolve, high-stakes topics (billing disputes, medical, legal), or negative sentiment that your policy says needs a person. ### Pair transfer with knowledge and routing Handoff works best with intelligent routing and a maintained knowledge base. Routing decides where the call should go; knowledge keeps answers accurate before a transfer is even needed. ### Test the unhappy path Do not only demo the happy booking flow. Call in and force an escalation. Confirm the human side sees useful context and that the caller does not have to repeat the entire story. ## Related capabilities [Multi-agent collaboration feature](https://ringbot.ai/features/multi-agent-collaboration)[Intelligent call routing](https://ringbot.ai/features/intelligent-routing)[Knowledge base documentation](https://ringbot.ai/docs/knowledge-base)[Getting started guide](https://ringbot.ai/docs/getting-started) ## FAQ ### Can RingBot transfer a live call to a human? Yes. Human escalation is a first-class part of multi-agent collaboration. When configured rules match — for example complexity, sentiment, or a caller request — the conversation can move to a human teammate while preserving what was already discussed. ### Is conversation context kept during handoff? Yes. Context preservation across agent-to-agent and agent-to-human transitions is a core platform capability. The receiving party should see the relevant conversation state rather than starting from a blank slate. ### Can one AI agent hand off to another AI agent? Yes. RingBot supports multi-agent collaboration: specialist consultation and seamless handoffs between agents, not only escalation to humans. ### When should I use transfer instead of a single mega-agent? Use transfer when different skills, departments, or compliance boundaries need separate ownership — for example booking versus billing, or AI triage versus a licensed human. Smaller, focused agents plus handoff are usually easier to maintain than one prompt that tries to do everything. ## Build agents that know when to pass the baton Start with a focused intake agent, add specialist paths, and define human escalation before you go live. [Get started](https://ringbot.ai/docs/getting-started)[All documentation](https://ringbot.ai/docs)