OpenAI has today launched OpenAI Presence, an enterprise\u00a0product\u00a0designed to help organizations deploy\u00a0reliable AI agents.\u00a0<\/p>\n
As agents take on more consequential customer and employee workflows, CX leaders will need to move beyond relying on controlled demonstrations and human review alone to judge whether systems are ready for production.\u00a0<\/p>\n
This launch comes as OpenAI and Hugging Face investigate a security incident in which advanced models exploited vulnerabilities to access information on Hugging Face\u2019s production infrastructure.\u00a0<\/p>\n
The Cost of Getting Deployment Wrong<\/p>\n
Today, enterprise AI agents are expected to carry out consequential tasks within\u00a0established boundaries,\u00a0consistently\u00a0completing\u00a0end-to-end workflows across customer service and internal operations. \u00a0<\/p>\n
When AI agents deployed to perform complex tasks or access sensitive information, they can introduce operational and regulatory risks that are not apparent in controlled AI prototypes.\u00a0<\/p>\n
This causes caution for organizations\u00a0expanding AI into high-value workflows,\u00a0as the supporting infrastructure required\u00a0for\u00a0autonomous agent safety has not always kept pace. \u00a0<\/p>\n
Businesses must\u00a0therefore\u00a0establish clear permissions and\u00a0governance frameworks, as without these safeguards, the\u00a0risk of\u00a0inconsistent customer experiences and unintended actions\u00a0increases, likely damaging trust.\u00a0<\/p>\n
In CX\u00a0environments\u00a0particularly,\u00a0organizations face increasing pressure to improve service availability while controlling operating costs. \u00a0<\/p>\n
In conversation with CX Today,\u00a0Sushil Kumar, CEO of Cyara,\u00a0argued that reducing AI hallucinations and maintaining reliable\u00a0CX requires governance that operates continuously alongside AI systems.\u00a0<\/p>\n
\u201cHuman oversight still matters, but it should function as a checkpoint rather than the entire CX assurance system,\u201d he argues. \u00a0<\/p>\n
\u201cEffective AI governance has to move at machine-speed, with automated validation, guardrails, and real-time testing.\u201d\u00a0<\/p>\n
Voice interactions\u00a0also add another layer of complexity because agents must understand requests\u00a0in natural language to\u00a0make decisions and respond in real time, particularly in highly regulated industries\u00a0where interactions may involve sensitive data, financial consequences, or timely\u00a0decisions. \u00a0<\/p>\n
Building for Controlled Scale<\/p>\n
The launch of OpenAI Presence\u00a0enables\u00a0organizations to\u00a0determine exactly what knowledge an AI\u00a0agent can access,\u00a0the\u00a0enterprise systems it may interact with,\u00a0and what actions it is authorized to perform, ensuring when\u00a0conversations or tasks should be escalated to a human. \u00a0<\/p>\n
This approach aims to ensure that AI agents operate within\u00a0defined boundaries, working alongside OpenAI\u2019s\u00a0enterprise customers through its Forward Deployed Engineers and selected systems integration partners to connect the agent with relevant business systems\u00a0and governance\u00a0policies\u00a0preparing\u00a0the deployment for production. \u00a0<\/p>\n
Presence includes tools that simulate real-world scenarios before deployment and continuously monitor production performance to identify weaknesses and opportunities for improvement.\u00a0<\/p>\n
Furthermore, the platform involves\u00a0Codex-powered improvement process,\u00a0designed to\u00a0analyze production data, investigate issues, and recommend updates to refine agent performance. \u00a0<\/p>\n
The performance of\u00a0Presence has reportedly shown\u00a0strong results for OpenAI\u2019s\u00a0English-language phone support service, where the AI agent resolves approximately 75% of inbound issues without human assistance,\u00a0with\u00a0its Codex-powered improvement process reducing\u00a0handoffs by 15%\u00a0over a 10-day period. \u00a0<\/p>\n
The Wider Safety Debate \u2013 Hugging Face<\/p>\n
OpenAI Presence\u00a0highlights the recent challenges surrounding the safety of deploying capable AI agents.\u00a0<\/p>\n