Joe Ashford

A Composer Who Wanted A Steadier Score

Music came first. Joe studied it at university and built a company writing music for television adverts and websites, a business that still runs on the side today. Commissions, though, arrive in bursts. They come in, they go quiet, and gaps in income are part of the deal. Something less ad hoc was the goal, and two conditions were non-negotiable: the work had to involve people, because Joe already knew how much there was to enjoy in talking to them, and it had to involve technology in some capacity.

A local recruitment agency happened to be hiring. Joe went in for a chat with one of the directors with modest expectations and came out with an offer as a junior consultant. That was twelve years ago. The aptitude showed up quickly, and so did the pattern that would define everything after it: a self-described geek who found a way to make a living inside the technology sector without ever writing the code.

Tech First, Everything Else Second

The early years were spent working with some of the largest names in consumer electronics, among them Samsung, Apple and Google. The people who left the deepest mark were not the ones handing over requisitions. They were the technical leaders and the managers running heavily research-led functions, the ones willing to open up their work and explain what they were actually building.

That generosity became a method. Rather than approaching researchers with a job description, Joe learned to arrive with genuine interest in the research itself, then trace how a specific piece of work fitted into a wider roadmap. It is not always easy for busy scientists to teach a recruiter their field. The ones who did handed over something no database provides: the ability to hold a real technical discussion with a PhD or a postdoc and be credible in it.

The habit carried straight through to quantum. Deep technology has been the constant across the whole career, and the standard is unforgiving. Requirements are stricter, candidate pools are smaller, and understanding the science is the price of entry.

An Aid, Not Gospel

Adoption came early. Joe was using large language models as soon as they became publicly available: feeding in job descriptions to see what came back, generating summaries, pressure-testing sourcing approaches. Today the tooling is woven into the working day, in use almost hourly for stakeholder management, scheduling, organizing hiring processes, note-taking, summarizing and turning that material into feedback for hiring managers. The verdict is unambiguous. It has made Joe better at the job.

Discipline lies in knowing the limits. Having recruited the teams that build these models, Joe understands the mechanics well enough to treat their output as an aid rather than gospel, and that awareness is what prevents over-reliance. A significant new piece of software is two weeks into trial, with eight people involved in the assessment, and it has proved better than expected. One line stays firm: a screening discussion is not something Joe would hand to an AI to host.

There is a symmetry to the shift that not everyone has noticed. Candidates are using the same tools, pairing their CVs with the job description to produce genuinely tailored applications, so the quality of what arrives has risen sharply. Better applications, in turn, make AI sourcing tools more effective at surfacing strong people. A double-edged sword, as Joe describes it, cutting in both directions at once.

Two Camps In One Profession

The industry splits into two camps, and the outlook is not identical for both. At the high-volume, lower-complexity end, where requirements are light and plenty of candidates could do the role, the technology can already find people, send outreach and manage a process along the way. Joe's read on that segment is blunt: much of the work will be wiped out, with a human still needed at some point in the sequence and far fewer of them employed to get there.

The other camp looks different. Executive hiring turns on nuance that resists pattern matching, because a CV carries keywords while a senior appointment depends on context and on understanding an individual beyond what is written down. The same holds at the most advanced technical end. At Oxford Quantum Circuits, sourcing tools genuinely help and Joe uses them, but with a PhD as the minimum bar, keyword matching accounts for only a fraction of the requirement. Complexity, on this reading, is what preserves the human role.

Advice: Plan First

For the year ahead, Joe's counsel to other talent leaders comes down to a single word: plan. The concern is specific and drawn from experience. Excitement about what AI might do has, in some businesses, moved straight to reducing headcount before anyone measured the true impact on the organization, and Joe has watched that go wrong more than once at previous employers.

The pattern is not confined to recruiting. Software engineering and graphic design get flagged as functions where impressive early results have prompted decisions that outran the evidence. Coding assistance is interesting and has produced good results, Joe allows, which is exactly why the temptation to act fast is so strong. The advice is deliberate rather than defensive: think hard before stripping bandwidth and capability out of a team, and do the planning before the decision rather than after it. That is the work Joe is doing right now, tooling up methodically while protecting the human judgment the hardest hires still require.