The unlovely support ticket has become one of artificial intelligence’s grand test cases. It contains intent, emotion, company knowledge, odd punctuation and, often, somebody’s dwindling patience. For Brad Murdoch, the chief executive of Deskpro, that makes it useful twice over: as a place where language models can do real work, and as a reminder that real work arrives with rules.
Murdoch’s pitch is more measured than the familiar promise of an overnight autonomous workforce. Let software summarize the long exchange. Let it draft the reply, identify intent, route the request and handle the repetitive question. Keep a person for the complicated case, particularly when empathy, creativity or judgment enters the room. Then make sure the system can operate in an environment that a security team will approve.
This is the thesis behind Deskpro Private, the company’s help desk offering for virtual private clouds, sovereign clouds, private data centers and even air-gapped deployments. Customers can connect a commercial, custom or private language model while keeping the help desk within their chosen perimeter. It is an infrastructure argument dressed in customer-service clothes, which may be why it fits Murdoch’s résumé so neatly.
“AI adoption is a journey, not a light switch.”Brad Murdoch, ITTech Pulse interview, 2026
He saw an old category with a new constraint
Murdoch first encountered Deskpro while working as an operating adviser for Elsewhere Partners, the Austin growth-equity firm that took a majority stake in the company. He had spent more than two decades with early-stage and growth-stage B2B software businesses. On paper, he later recalled, help desk software looked saturated. The shelves already groaned with recognizable names.
What changed his reading was Deskpro’s flexibility. The company, founded in 2001 in an Oxford University dorm room, had grown with both cloud and self-hosted customers. That dual history looked unfashionable during the years when every software story wanted to become pure multi-tenant SaaS. In the age of AI, it became a strategic option. Regulated organizations wanted the productivity gains of language models but could not casually send sensitive support records into a public-cloud service.
In June 2024, Elsewhere Partners invested $25 million. Murdoch joined as CEO, Deskpro opened a global headquarters in Austin, and founder Christopher Padfield shifted his attention toward product strategy as chief product officer. The arrangement paired a founder-led product history with an operator whose career had passed through nearly every room in a software company.
A career assembled from handoffs
Murdoch’s public career reads less like a straight ladder than a series of handoffs between functions. At Wind River, he led services. At JBoss, he held responsibility for services and operations before Red Hat acquired the open-source middleware company in 2006. At OpenSpan, he worked across products and strategy. He co-founded mobile application management company Nukona and served as its chief marketing officer before Symantec bought it in 2012.
Next came the chief marketing officer role at Framehawk, acquired by Citrix in 2014, and then application-security company Prevoty, later acquired by Imperva. At Lightbend he moved through corporate development and product strategy into an executive vice president role. Along the way he became a board member and adviser as well as an operator. His public profile lists the University of Glasgow, without specifying a degree or dates.
Selected stops in a cross-functional software career
That range matters in enterprise software because a feature rarely remains a feature. It becomes a deployment choice, a procurement sequence, a security review, a migration plan and a new habit for the person expected to use it. Murdoch’s current argument does not treat those steps as debris around the product. They are the product’s path into production.
The support team changed its mind
One of Murdoch’s more revealing examples comes from Deskpro’s own support operation. When the company introduced AI capabilities, he said, the team initially pushed back because employees worried the tools might replace them. Then the tools began drafting replies, summarizing ticket histories, helping create knowledge-base articles and reducing repetitive work. The support team’s position changed. According to Murdoch, they eventually refused to let the company turn the features off.
The anecdote is modest, even domestic, beside the industrial language of agentic transformation. It also contains a better adoption playbook. People trust software when it removes friction they recognize and leaves them with agency. A drafted answer can be corrected. A summary can be checked. A routine question can be deflected while the difficult one still reaches a human.
“The best use of AI is to assist, not replace.”Brad Murdoch on Deskpro’s AI pricing, 2026
That philosophy reaches the price sheet. In August 2026 Murdoch argued against charging customers for each AI “resolution,” a unit whose definition can vary by vendor. Deskpro favors token-based costs and a bring-your-own-AI option. His reasoning is that support teams should use models broadly as assistants, not optimize the system around preventing human involvement. Economics, in this telling, can quietly shape product behavior.
Security is the route, not the roadblock
Deskpro’s 2025 research of more than 220 support, IT and HR professionals found a sharp split. The company reported AI adoption in support operations at 92 percent among technology companies and 58 percent across regulated industries. Seventy-one percent of support leaders in regulated sectors cited compliance as the primary reason they could not deploy AI, while 64 percent said they would deploy only with full control over data and infrastructure.
Reported adoption of AI in support operations. Deskpro surveyed more than 220 professionals across support, IT and HR functions in 2025.
Murdoch’s phrase for the split is crisp: the adoption gap is an architecture problem. A bank, government agency or defense contractor may understand the potential of AI perfectly well. The obstacle is where information travels, who can inspect it and which jurisdiction applies. Treating those requirements as timidity misses the buyer’s actual job.
One product, four controlled routes
Deskpro Private launched in November 2025 with the ability to connect commercial, custom or private models inside those environments. In March 2026, the company announced support as a partner for the AWS European Sovereign Cloud. In June, Deskpro Private became a CODiE Awards finalist for customer service software. These are company milestones, not proof that the thesis has won. They do show the direction holding steady.
When a help desk becomes a verb
Murdoch’s larger product idea begins with a small annoyance: nobody particularly wants to visit a help desk. They want an answer. In his example, an employee asks an HR question in Slack. The system recognizes the intent and decides whether an AI agent can answer from approved information. If the question is too complex, it routes the request to a person who can respond in the same channel. The employee does not retype the story into a portal.
He describes this as turning the help desk from a product into a function. Slack is only one doorway. Microsoft Teams, email, chat, social channels and voice can all feed the same contextual service. The interface recedes while the organizational memory and routing logic remain. It is less a dramatic reinvention than a good piece of plumbing, which is often what people mean when technology finally works.
Murdoch’s career gives the idea an extra resonance. He has repeatedly worked at the seams: services and product, marketing and strategy, investor and operator, cloud convenience and private control. Deskpro is now betting that the neglected seam in AI support sits between capability and permission. The language model can compose a reply. The more consequential question is whether it belongs in the room at all.
“The help desk is being redefined: it now becomes a function rather than a product.”Brad Murdoch, Deskpro interview, 2026
There is no theatre in asking a security team to sign off. There is paperwork, architecture, negotiation and the occasional grim spreadsheet. Yet for enterprise software, that is where aspiration becomes use. Murdoch’s wager is that the next useful chapter of support AI will not be written by removing every constraint. It will be written by designing carefully enough to live with them.