OpenAI announced Presence on July 22, 2026: a managed platform for deploying real-time voice and chat agents that handle customer support, sales, and internal workflows under company-defined policies and guardrails.

The facts: Presence is a deployment layer that sits on top of OpenAI's models, not a model itself. It isn't self-serve — rollouts are led by OpenAI's own Forward Deployed Engineers alongside select systems integrators, aimed at healthcare, financial services, and other regulated or complex operations. Pricing and contract terms haven't been published. OpenAI says its own English-language support line, running on Presence, resolves 75% of inbound calls with no human involved — a strong number, though it's their number, about their line, not an independent audit.

Automation Squad's take: for most execs reading this, Presence itself isn't the actionable part — nobody's getting a call from OpenAI's deployment team this quarter. What is actionable is the question it raises: how much of your own phone or chat support is repetitive enough to automate, and do you actually know that number? That's answerable today, with the AI tools you already have, well before any enterprise voice platform is worth a serious look.

Run this now: pull a rough log of your last 100 customer support calls or tickets — even just subject lines — and paste the list into ChatGPT or Claude with this: "Here's a list of our recent support contacts. Group these into categories, estimate what percentage of each category is a repeatable, rules-based answer versus something that genuinely needs a human judgment call, and tell me which category would be the safest place to pilot an AI agent first." That answer tells you whether Presence-style automation is even worth chasing yet.