Every hiring conversation I have at the moment eventually lands on the same question: what do we actually call this role, and what should we pay for it?
That question did not exist eighteen months ago for most of the titles now appearing on client briefs. AI has created a wave of new roles faster than the market has been able to price them, and that gap between demand and data is exactly where H2 2026 gets interesting.
I have split this into three groups (updated from two). The first is AI roles where we now have real benchmark data, built from live placement activity across our platform. The second is roles we are seeing named in briefs and job descriptions, where demand is clearly building but the salary market has not settled yet. The third is one role in particular that deserves its own section, because it is moving faster than either of the other two lists.
AI roles with live benchmarks now
These titles have moved from emerging to established over the past year. Salary growth across AI, machine learning and senior data roles has outperformed the wider market by more than 12% over the last twelve months, and it shows in the numbers below.
| Role | Permanent (low–high) | Contract inside IR35 | Contract outside IR35 |
| Head of AI | £120,000–£200,000 | £850–£1,000 | £800–£950 |
| AI Governance Lead | £90,000–£110,000 | £750–£850 | £650–£750 |
| AI Architect | £110,000–£150,000 | £850–£1,100 | £750–£1,000 |
| AI Product Manager | £90,000–£130,000 | £850–£1,000 | £700–£900 |
| AI Engineer (mid to principal) | £80,000–£150,000 | £800–£900 | £700–£800 |
| MLOps Engineer | £80,000–£120,000 | £750–£850 | £700–£800 |
| AI Business Analyst | £60,000–£90,000 | £650–£750 | £550–£650 |
A few patterns stand out here. AI Governance Lead has moved fastest from concept to benchmark, which tells you everything about how quickly organisations are treating AI oversight as a hiring priority rather than a policy document. Head of AI is now commanding CDO-adjacent salaries, appearing across retail, healthcare, manufacturing and the public sector, not just technology and financial services.
AI roles still forming
These are the titles turning up in briefs where the demand is real but the salary benchmarks are still settling. If you are hiring for one of these now, expect to negotiate on first principles rather than a published range.
- AI Product Owner – sitting alongside AI Product Manager, but with a narrower delivery focus. We are seeing this appear as organisations split AI strategy from AI delivery.
- Agentic Operations Manager – a direct response to agentic AI moving from concept to operational planning. This role owns the workflow orchestration and human-in-the-loop design that goal-directed AI systems require.
- Human-in-the-Loop Designer – governance-adjacent, focused on where and how human oversight sits inside automated decision-making. Expect this to grow fastest in regulated sectors.
- Finance AI Lead – appearing inside finance transformation teams as AI-enabled analytics and forecasting move from pilot to core process.
- ESG Reporting Lead – not purely an AI role, but increasingly AI-adjacent as organisations use AI-driven analytics to meet reporting requirements.
- Marketing Operations Lead, AI-enabled – a CRM and marketing operations role reshaped by AI-enabled customer workflows and journey optimisation.
- Salesforce AI Specialist – combining platform configuration with AI capability, a profile that is genuinely hard to find right now.
The one to really watch: forward deployed and applied AI engineers
There is a third pattern worth pulling out on its own, because it is moving faster than anything else covered here: forward deployed engineering.
The title started at Palantir, and has since been adopted, with slight variations, by most of the major AI labs. OpenAI hires Forward Deployed Engineers. Anthropic calls the same function Applied AI Engineers. The job is the same either way: embed directly with a customer, learn their workflow, and build the system that gets an AI model from proof of concept into something running in production on their data, inside their constraints.
The demand curve on this is genuinely unusual. Research from Christian & Timbers, based on interviews with more than 250 C-suite hiring executives, projects demand for forward deployed engineers to surge by 2,100% by the end of 2026. At the start of the year, only 5 to 10% of companies were planning to hire for the role, mostly for small pilots. By the end of the second quarter, that had jumped to 70%, with some of the largest consulting and services firms reporting a need to increase forward deployed headcount tenfold, building full teams of 20 to 100 people. Indeed has seen job postings for the role jump 729% year on year.
Supply has not moved anywhere near as fast. Estimates put the number of genuinely qualified specialists at only a couple of thousand globally, and the UK slice of that is small. Roles are appearing in London, and to a lesser extent Manchester, Glasgow and Belfast, but the profile itself, someone who can write production code and also run a boardroom conversation with a client’s CTO, is rare wherever you look for it. Recruiters working this market describe it as the tightest senior tech search they’re currently running, with six to ten week fill times the norm rather than the exception.
Compensation reflects the scarcity. UK-based roles at growth-stage AI companies are advertising in the £60,000 to £110,000 range plus equity, while forward deployed and applied AI engineering roles at the frontier labs in the US are clearing significantly higher. That gap alone tells you where the competitive pressure sits.
What this means for organisations hiring into this space
There simply are not enough UK-based forward deployed or applied AI engineers to hire your way through this in the traditional sense. Organisations serious about this capability need to think more creatively than a standard permanent search: widening the net to include international or remote-first candidates, building the profile internally from strong generalist engineers with client-facing potential, or borrowing the capability through interim and contract delivery while a longer-term plan comes together. This is exactly the kind of role where the hire, build, borrow framework earns its keep, because a hire-only strategy is going to run into a supply wall fairly quickly.
Why the benchmark gap matters
Adoption is a metric. Capability is a strategy. That applies to hiring too. If you wait for a published salary benchmark before you write the job description, you will be negotiating against organisations who moved six months ago.
My advice for any of the roles in the second list: benchmark against adjacent established roles, budget for a premium given constrained supply, and be honest with candidates that the title and scope may still evolve as the market catches up. Candidates taking these roles right now are often defining them as much as filling them.
What this means for your workforce plan
Whether you are hiring, building or borrowing this capability, the same principle applies across both lists. Specialist AI capability is constrained, and that constraint is not easing through H2. Organisations that move early on governance, agentic operations and AI-enabled functional roles will be negotiating from a position of strength that gets harder to reach every quarter this goes unaddressed.
Our H2 2026 Workforce Insights Report has full salary benchmarks for every role in the technology market, alongside the roles to watch across every practice area. If you are hiring for one of the roles still forming, get in touch. This is exactly the kind of brief my team works on every week.