OpenAI Positions HP as a Blueprint for AI as an Operating Layer

HP has announced a strategic partnership with OpenAI to deploy its Frontier platform across customer-facing experiences and internal operations. Partnerships like this have become routine in enterprise AI, but one line from the announcement deserves more attention than the deal itself. Denise Dresser, OpenAI's chief revenue officer, said HP "is showing what enterprise transformation looks like when AI becomes an operating layer - connected to the systems and workflows where work already happens."

OpenAI is beginning to describe AI less as an assistant or even an agent, and more as something that connects enterprise systems and workflows together. HP looks set to be the company that puts that idea into practice first.

HP's planned use of Frontier spans customer- and partner-facing solutions, customer telemetry insights and reporting through its WXP platform, employee productivity, and software development. The partnership follows pilots that began when HP started testing Frontier in February 2026, giving the company roughly four months of evaluation before committing to wider deployment.

None of these use cases is new on its own. What's different is having one AI platform sit behind customer interactions, employee productivity and software development at the same time, rather than separate tools running in separate departments with no shared context between them.

The vocabulary of enterprise AI has shifted a few times in the past three years. Assistants were built to help individual employees get through their day. Agents went further, automating workflows and handling multi-step tasks without much human input. What's emerging is something with a wider remit. AI is now meant to coordinate work across people, systems and entire business functions, not just complete tasks within them.

HP's own framing, that "AI is becoming a new layer for how work gets done across the company" chimes with that description. AI is no longer being sold as another application sitting next to existing software, but as the base layer infrastructure.

CX-focused AI has traditionally meant chatbots, agent assist, call summarisation and workforce optimisation, each largely self-contained. An operating layer doesn't work that way. It depends on AI connecting customer data, enterprise knowledge, engineering teams, back-office operations, sales and marketing, the various pieces that already make up the modern CX AI stack, so that customer experience becomes one output of a wider system rather than a project run on its own.

For CX leaders, that pushes the conversation away from buying individual point solutions and towards how AI gets woven into the business as a whole, with customer experience as one of several things it's expected to improve.

HP and OpenAI aren't alone here. Research cited within Salesforce's own coverage found that German enterprises are already redesigning how they use its platform and "making it an operating layer for generative AI and agent-based work" across sales, service, IT and industry operations. Verint is working towards the same end but from a different angle, framing its Agent Factory platform around managing mixed human and AI workforces rather than systems integration. Microsoft has built a coherence argument around its own stack, positioning Copilot Studio and Dataverse as the shared layer behind its contact centre, sales and customer insights products. Meanwhile, ServiceNow has been the most upfront, calling itself "the AI control tower for business reinvention" as it extends governance across every AI system, agent and workflow it can reach.

The terminology differs across all four, but the underlying architecture they're each building towards is essentially the same.

Part of the reason is that the foundation models themselves have become less of a differentiator, as gains there have spread fairly evenly across vendors. The harder, more valuable problem now sits in the layer connecting data, coordinating workflows, managing permissions, and holding onto organisational context that a model alone can't supply. That would explain why so many vendors are pushing into the same territory at once, while moulding it to their differing platforms.

For much of the past year, enterprise software vendors have argued that AI's future lies in coordinating work across the organisation rather than simply assisting individual users. HP's partnership suggests OpenAI is now telling the same story. The language may differ from vendor to vendor, but the direction of travel is becoming increasingly consistent.