RingCX puts AI and human agents on one QA rubric

RingCentral's RingCX now scores AI agents against the same quality rubric as human agents. Announced October 28 at ICMI 2026 in Orlando, the update puts AI agent calls in the same live supervisor view as human calls, hands calls between AI and humans in both directions with context kept, and forecasts staffing across both. The AI agents covered are RingCentral's own AI Receptionist (AIR) and AIR Pro. RingCX is also now available on mobile and inside Microsoft Teams.

What actually happened

Per RingCentral's press release:

• Two-way handoff: calls move from AI to human and back without losing context.
• Live supervision: AI interactions show up next to human calls, so a supervisor can step in on either.
• Same QA rubric: every AI interaction is scored against the rubric used for human agents.
• Journey analytics and forecasting: track the full AI-to-human path, and plan staff with AI capacity counted.
• AI Onboarding: an admin describes a setup in plain language; RingCentral's AI builds it through the RingCX APIs and runs a simulated routing test before go-live.

What the release does not include: pricing, a GA date for the AI features (they were "showcased"), or any performance numbers. Treat it as a direction, not a spec sheet.

Why one rubric for AI and human agents matters for your business

Most small teams that add an AI receptionist never grade it. They check that it picks up and books appointments, then move on. Meanwhile a human rep who misquoted a price or lost a caller would get coached the same week. The AI gets no review at all.

Scoring both against one rubric fixes that. It also gives you the only honest answer to "is the AI working?": its scores next to your people's, on the same calls.

You do not need RingCX to do this. Here is the version we set up for clients:

1. Write the rubric in five lines. Greeting, correct answer, correct next step, tone, clean handoff.
2. Pull 20 AI calls and 20 human calls a week. Score them blind if you can.
3. Grade the handoff, not just the bot. Did the human have to ask the caller to repeat everything? That is a failed transfer.
4. Set a floor. If the AI scores below your newest hire for two weeks, narrow what it handles.

And for AI Onboarding: a simulated test before go-live is good. Still call your own line before you switch routing.

Sources: RingCentral press release, Business Wire.