OpenAI brings real-time interruptible voice AI to enterprise workspaces and launches Presence for customer-facing agents

OpenAI on July 23, 2026, launched GPT-Live inside ChatGPT desktop work surfaces for enterprise workspaces, enabling real-time interruptible voice interaction with tasks and agents, and separately introduced Presence, a managed enterprise product for voice and chat agents in customer support and other workflows. Presence, deployed through Forward Deployed Engineers and systems integrators, powers OpenAI's own support line at 1-888-GPT-0090, resolving 75% of inbound issues without human assistance and reducing human handoffs by 15 percentage points in 10 days via a Codex-powered improvement loop.

OpenAI's voice push now has two fronts: GPT-Live for workers who want to talk their way through tasks, and Presence for companies that want AI agents handling real support calls. OpenAI's enterprise voice rollout landed on July 23, 2026, and it's not just a nicer microphone button. It puts GPT-Live inside ChatGPT's desktop work surfaces, so you can speak to ChatGPT while it starts tasks, checks progress, interrupts an agent, or coordinates several Work and Codex threads at once. That changes the job of voice. It's no longer only a way to ask a question out loud. The pricing is concrete enough to make finance teams pay attention. OpenAI's Help Center says Business workspaces include one hour of Voice in Chat, with extra use billed at five credits per minute. Voice in Work and Codex costs roughly six credits per connected minute where flexible pricing applies, while the underlying Work or Codex tasks still draw from the normal shared usage pool. So yes, you can tell an agent to open a new thread, chase a bug or keep a project moving while you talk. You'll also need to watch the meter. GPT-Live-1 and GPT-Live-1 mini launched on July 8 as full-duplex voice models. Full-duplex is the important part. The system can listen and speak at the same time instead of waiting for a clean pause before responding. You can interrupt it mid-sentence. It can keep listening while it delegates harder questions to another model in the background. According to OpenAI's own launch post, more than 150 million people use ChatGPT through Voice and Dictation each week, and GPT-Live is now powering the improved ChatGPT Voice experience. That's a large base to shift. TechCrunch's coverage of the July 8 launch put the change plainly: OpenAI was replacing its older Advanced Voice Mode experience with GPT-Live-1 mini by default for free users, while paid users would get access to the larger GPT-Live-1 model. That detail matters because this is an architecture change, not a skin on the old product. The old voice flow felt like a turn-based meeting. This one is built to deal with interruption and background work - pauses included. Presence is the heavier product Presence arrived separately on July 22. Don't lump it in with ordinary ChatGPT Voice. Presence is OpenAI's managed enterprise product for voice and chat agents across customer support, billing issues, IT help desks, procurement, HR, claims and sales workflows. It's not self-serve. OpenAI says deployments are handled through its Forward Deployed Engineers and select systems integrators, with access limited to eligible enterprise customers. That service model tells you who OpenAI is chasing. Salesforce has Agentforce. Microsoft has Copilot Studio. Contact-center AI startups have been selling voice agents into support teams for years. OpenAI is now selling against all of them with a product that wraps models in policies, permissions, simulations, evaluations, guardrails and escalation rules. It's less glamorous than a new model name. It's also the part enterprises actually buy. The strongest proof point is OpenAI's own support line. According to the company's Presence launch page, Presence powers OpenAI's English-language phone support channel at 1-888-GPT-0090, met or exceeded the company's frontline human-support quality benchmarks within weeks, and now resolves 75% of inbound issues without human assistance. Keep the caution attached to that number. It's self-reported. It hasn't been independently audited. Still, it's a real operating claim from a real queue, and that's more useful than the usual demo clip. OpenAI also says Presence's Codex-powered improvement loop reduced human handoffs by 15 percentage points in 10 days. That's the detail enterprise buyers should sit with. A voice agent that works on launch day is useful. A voice agent that keeps adapting as policies, products and customer behaviour change is the harder promise. Here's the thing: customer operations break in the gaps. Refund rules change, account screens move - and suddenly the assistant that looked good in testing starts making a mess. The buyer still has to do the work Presence doesn't remove the hard work from the company buying it. It makes that work visible. You still have to decide what an agent can see, what it can do, when it needs approval, and when a person takes over. OpenAI's product page names BBVA, SoftBank and IAG as enterprises working with Presence, including BBVA's exploration of voice support for everyday banking in Mexico and SoftBank's testing of Japanese-language customer conversations. Those are exactly the places where vague automation talk meets regulated, high-pressure reality. The workspace rollout has a different audience. It's for people who already spend their day inside documents, code, browser tabs and internal systems, and want voice to steer work without stopping to type every instruction. On desktop, OpenAI says Voice in Work and Codex can use the tools and permissions available in the selected experience, including computer context where enabled. That's powerful. It's also bounded. Voice can coordinate agents, but it still lives inside the permissions and surfaces OpenAI exposes. Costs will shape adoption. Five or six credits per minute sounds manageable when one person tests it for a short call. It looks different when a support team, IT desk or engineering group starts leaving voice sessions open during real work. Microsoft and Salesforce can fold agent features into large enterprise contracts. OpenAI is asking buyers to think about usage more directly. The open question isn't whether voice AI can sound more natural. GPT-Live already moves that forward. The real question is whether companies will trust it with the messy work that touches customers and internal accounts - the stuff where errors have consequences. Presence is OpenAI's answer: bring engineers, policies and guardrails, then prove it in production. That's not a toy launch. It's enterprise software, with all the boring obligations that come with it. Also read: GlossGenius rebrands as Genius AI and closes a $44M Series D at a $1.15 billion valuation https://startupfortune.com/glossgenius-rebrands-as-genius-ai-and-closes-a-44m-series-d-at-a-115-billion-valuation/ • Verizon signs a dark fiber deal with Google worth over $1 billion as AI infrastructure demand reshapes telecom https://startupfortune.com/verizon-signs-a-dark-fiber-deal-with-google-worth-over-1-billion-as-ai-infrastructure-demand-reshapes-telecom/ • Meshy just closed the biggest AI-3D funding round ever and half of Big Tech is already a customer https://startupfortune.com/meshy-just-closed-the-biggest-ai-3d-funding-round-ever-and-half-of-big-tech-is-already-a-customer/