Report on how tech jobs are changing, what's evolving, what's new, and what's fading across product, engineering, and operations. Plus an interactive map of 24 roles across 2026, 2028 and 2030.
Walmart created a new top role to lead AI, product, and design company-wide. The title was EVP of AI Acceleration — a role that did not exist a year earlier.
Amazon's CEO told employees that AI would reduce the company's corporate workforce.
Tesco partnered with Mistral for three years and launched a joint AI lab in London.
Walmart, Amazon, and Tesco made different AI moves: new roles, fewer jobs, new partnerships. Same signal: AI is redrawing retail IT organisational charts.
This briefing is not about whether AI is coming for IT work. It is about what shape an IT organisation takes when ninety percent of its engineers use AI tools by 2028, and three out of every four IT jobs are augmented by AI by 2030. Both figures: Gartner.
The brick building sits along the water at Houthaven, in the west of Amsterdam. Tall glass windows. The canal outside is calm. Hanna walks through the open atrium with a coffee and finds her team in a small meeting room: two product owners, two designers, two engineers, one data analyst. A big whiteboard along the short end. She sets down her coffee and the team begins.
They are prioritising three features for the next sprint. A smarter shift planner for store associates. A reworked break-room scheduler. A pilot for an assistant that helps colleagues swap shifts without asking a manager. Hanna runs the prioritisation. She sketches a 2x2 on the board, the way she has done many times before.
Her laptop is open on the table. A Slack thread is going. Someone in Helsingborg used Copilot to draft three of the user stories overnight. They are good. Too good. The team is quietly uneasy in a way nobody mentions.
Hanna closes the laptop and looks out at the water. Her parents in Birmingham still ask what she does. She still tells them she is a product owner. The sentence is true.
She is not yet sure for how much longer.
Team size compresses. Team scope expands. The product owner of 2026 still exists in 2028. She just owns more, with fewer people around her, and most of the work is reviewed rather than written.
of enterprise software engineers will use AI code assistants by 2028.
Up from less than 14% at the start of 2024.
Gartner · October 2024 →Median expected workforce decrease in functional teams over the next twelve months, in a McKinsey survey of organisations using generative AI.
Self-reported by leaders. Lagging indicator. Direction of travel is clear.
McKinsey · March 2025 →Salesforce customer support roles in 2025. Headcount fell from 9,000 to 5,000.
The CEO said on the record: I need less heads.
CNBC · September 2025 →The honest counter-point. Klarna built the most public AI-first customer service operation in 2024 and then publicly reversed course in May 2025, rehiring humans because customer satisfaction on complex cases collapsed. The lesson the rest of the market has now absorbed: hybrid is the steady state. AI handles volume. Humans handle nuance, brand, and edge cases.
The under-told story. The compression is not really about senior product managers. It is about junior ones. Stanford's Digital Economy Lab found a sixteen percent relative employment decline among workers aged twenty-two to twenty-five in the most AI-exposed occupations. The career ladder is losing its bottom rungs faster than its top ones.
Hanna gets to the office before the others. The light over the canal is soft and grey-blue. She stops by the window for a moment before sitting down.
On her screen, forty-seven user stories generated since she went home yesterday. (User stories live in the product backlog and become the work the team prioritises for upcoming sprints.) She reads them carefully. Twenty-eight she approves. Eleven she rewrites. Eight she rejects, with a written reason, because the agents learn from rejections.
The whole loop has changed pace. Two years ago her team ran the company sprint cycle the way most product teams did: explore and prototype, test, develop, launch to markets, measure and plan the next sprint. Now most of those phases run faster than humans can review them. The biggest shift, the one nobody talked about loudly: developers became the fast ones. Designers used to be the quick prototypers waiting for engineering. By 2028, dev outruns design.
Two of her designers left in 2027. They were not replaced by designers. They were replaced by a single role her organisation now calls AI Output Reviewer, which is not quite a designer and not quite an engineer and not really a product owner. Hanna does not yet have a clean way to describe this person to anyone outside the building.
A message from her mother. Beta, what is it you actually do these days? Hanna stares at the question for a long time. She types: I lead product development. She deletes it. She types: I make sure the work the machines do is the work we meant.
She sends that one. It is closer.
Specialised operations roles change function rather than disappear. The "process administrator" version of each role thins out. The "AI governance owner" version of each role grows. The layer above and around (managers, ops, assistants, comms) bends harder than the product teams themselves.
of organisations now have a Product Ops function.
43% of practitioners say AI and automation will define the next three years of the role.
Productboard · May 2025 →of enterprise applications will feature task-specific AI agents by 2026.
Up from less than 5% in 2025. Eight times growth in one year.
Gartner · August 2025 →UX design and UX research postings since the start of 2022.
The functions that survive take ownership of evaluating AI-generated work.
Indeed labour-market data · 2025 →of organisations will use AI to flatten their structure by 2026, eliminating more than half of current middle management positions.
One in five. The shift is structural, not cosmetic.
Gartner · October 2024 →What appears in their place. Microsoft's 2025 Work Trend Index found 28% of managers are already considering hiring AI Workforce Managers. Eightfold and Deloitte both name AI Agent Orchestration Specialist as the most important hire of 2026. Cognizant publicly committed to hiring a thousand "Context Engineers" in August 2025. Some of these titles will not survive. Some will. The point is not the names. The point is that the org chart now needs new boxes.
The office at Houthaven looks the same. The water outside, the brick walls, the tall glass. Hanna sits at a desk by the window. Twelve agents work for her, three humans on her team. An architect. A governance specialist. An Intelligence Resources lead.
She no longer runs prioritisation workshops. On Monday mornings she translates business outcomes into agent instructions. The rest of the week she handles trust gaps: stepping in when colleagues do not trust what the agents have produced, helping people decide whether to accept the work, override it, or ask for a different cut. The translation work takes an hour. The trust work takes the week.
A young woman calls her on Teams. New hire, six weeks in. Brilliant. Confused. She asks: How did you know what your job actually was, when everything kept changing?
Hanna takes a breath. Outside the window, a container ship slides past, getting ready to leave the harbour. She thinks about the version of herself in that small meeting room in 2026, choosing between three features.
I didn't, she says. I just kept asking what was still mine to decide. And then I did that part well.
The call ends. Hanna sits with the empty screen for a moment before her next meeting.
Twenty-four roles, three time horizons. Click any row to read what happens to that role in 2026, 2028 and 2030. Yellow blocks are roles that did not exist before. Faded dashed blocks are roles on the way out.
| Role | 2026 | 2028 | 2030 |
|---|---|---|---|
| Product teams | |||
Product Owner Product team |
|||
2026 Continues Owns roadmap, prioritisation, stakeholder alignment. Uses AI for research and drafting but holds the pen. 2028 Mutates Owns more outcomes with fewer people. Reviews work generated by agents. The "writing" part of the product owner role compresses to almost nothing. 2030 Mutates further Title often changes to Outcome Owner. Most of the role is translation between business intent and agent instructions, plus closing the trust gap when humans inside the org do not trust the work an agent has produced. | |||
Product Designer Product team |
|||
2026 Continues Designs flows, components, design systems. Uses AI as a faster sketchbook. 2028 Mutates Most "production design" is AI-generated. Designer's job becomes setting design intent, evaluating output, and maintaining the system the AI generates against. 2030 Shrinks Headcount falls noticeably. The senior designer survives as art director and taste authority. Junior designer roles thin out the most. | |||
UX Researcher Product team |
|||
2026 Continues Runs interviews, synthesises insights, owns participant access. 2028 Shrinks AI handles transcription, coding, and first-pass synthesis. Headcount contraction follows. The role survives as research strategist plus AI-output validator. 2030 Shrinks further Centralised research team becomes small and senior. "People who do research" is now distributed across product and ops. | |||
Software Engineer Product team |
|||
2026 Continues Writes code, reviews code, owns architecture decisions. AI assists, does not replace. 2028 Mutates Per Gartner, 90% use AI code assistants by 2028. Junior pipeline narrows. The engineer's day is mostly review, integration and orchestration. 2030 Mutates further Headcount roughly steady at senior level. Title often becomes Platform Engineer or AI Systems Engineer. Pure feature implementation is largely automated. | |||
Data Analyst Product team |
|||
2026 Continues Builds dashboards, writes SQL, runs ad hoc analysis for product teams. 2028 Mutates Self-serve analytics through natural-language interfaces takes most ad hoc work. The analyst becomes data quality steward and metric-system architect. 2030 Mutates further Closer to "Decision Support Engineer." Owns the trustworthiness of data feeding agent decisions, not just dashboards. | |||
QA Engineer Product team |
|||
2026 Continues Manual and automated test, exploratory testing, regression coverage. 2028 Shrinks Test generation, regression, exploratory and accessibility checks largely covered by AI. Headcount falls fast. 2030 Disappears as a discrete role QA is absorbed into platform engineering and AI evaluation pipelines. The discipline survives, the standalone role rarely does. | |||
| Specialised operations | |||
Product Ops Specialised ops |
|||
2026 Continues Roadmap admin, tool stack, release coordination, customer-feedback aggregation. 2028 Mutates Becomes the function that operationalises AI inside product. Owns prompt libraries, agent observability, eval-set management. 2030 Mutates further Closer to AI Operations Lead. Strategic seat, not admin. | |||
Tech Ops / Platform Engineering Specialised ops |
|||
2026 Continues Standalone DevOps still recognisable. Internal developer platform investments accelerate. 2028 Mutates DevOps as a separate discipline absorbed into platform engineering. Owns AI tooling, eval pipelines, agent-supervised incident response. 2030 Continues at higher level One of the most stable functions, but the work itself is unrecognisable from 2026. Platform engineering is now AI-platform engineering. | |||
Design Ops Specialised ops |
|||
2026 Continues Manages design system, tools, hiring, process for the design org. 2028 Shrinks Process-admin version contracts. Standalone Design Ops headcount falls. 2030 Mutates Survives as guardian of AI-generated design output: brand-safety guardrails, evaluating how well an AI system performs against design intent, and calibrating output so it consistently meets quality standards. | |||
Research Ops Specialised ops |
|||
2026 Continues Recruitment, scheduling, repository, tool stewardship for research. 2028 Shrinks Transcription, coding and tagging are now AI-native. Smaller team. User Interviews 2025 ReOps survey: 12 of 21 specialists report AI has greatly transformed their roles. 2030 Mutates Tool-governance and AI-output-validation function. Smaller, more senior, more strategic. | |||
| Supporting and operational | |||
Executive Management Supporting |
|||
2026 Continues The leadership group that runs the IT organisation: heads of product departments, plus heads of supporting functions (HR, communications, business steering, finance ops). 2028 Mutates The group composition shifts. Heads of supporting functions whose work has been heavily automated lose their seat. Heads of newly emerging functions (Intelligence Resources, AI Operations) gain one. 2030 Mutates further An "AI triad" emerges in many enterprises: CEO, CTO, and Head of Business Operations as the operating-model owners. Other exec functions report into or around the triad rather than as parallel peers. | |||
HR / People & Culture Supporting |
|||
2026 Continues Hiring, onboarding, performance, talent development, employee relations. 2028 Mutates Owns AI-literacy programmes, agent-supervision training, hybrid-team workforce planning. Operational HR (admin, benefits, payroll) is largely automated. 2030 Mutates further Splits or rebrands. The strategic half often becomes "Intelligence Resources." The remaining People & Culture function focuses on what is genuinely human: belonging, motivation, navigating fast change. | |||
Engineering Manager Supporting |
|||
2026 Continues Spans of control around 6 to 10. People management, technical leadership, delivery. 2028 Mutates Spans expand. Some teams report 20-plus reports including humans and agents. Microsoft 2025: 28% of managers considering hiring AI Workforce Managers. 2030 Shrinks Layer thins. Many functions go from three management layers to two. | |||
Programme / Project Manager Supporting |
|||
2026 Continues Gantt charts, dependencies, status reports, risk register, stand-up synthesis. 2028 Shrinks Most coordination work goes to agents. Headcount falls fastest in tech-heavy orgs. 2030 Mutates Survives as outcome owner on politically sensitive programmes (M&A, restructures, regulator-facing work) where stakeholder navigation still requires a human. | |||
Business Operations Supporting |
|||
2026 Continues KPI decks, variance analysis, reorg modelling, cross-functional projects. 2028 Mutates Excel-and-PowerPoint version contracts. The function moves up the value chain into workforce-architecture decisions: which workflows go to agents, which stay human. 2030 Mutates further Often sits inside an AI Triad with CEO and CTO. Business Ops becomes the operating model owner. | |||
Operational Strategist Supporting |
|||
2026 Continues Strategy, scenarios, prioritisation, board-level analysis. 2028 Continues, AI-augmented Tooling shift, not headcount shift. Strategists with AI fluency outperform. 2030 Continues One of the most resilient roles in the org. Judgement under uncertainty does not commoditise. | |||
Executive Assistant Supporting |
|||
2026 Continues Scheduling, expenses, travel, inbox triage, meeting prep. 2028 Shrinks at junior level WEF Future of Jobs 2025 names admin assistants among the fastest-declining occupations through 2030. Junior EA work is mostly automated. 2030 Shrinks further, with a senior tail Senior EAs survive as Chief-of-Staff-lite figures: judgement, political navigation, executive proxy. | |||
Internal Comms Supporting |
|||
2026 Continues Newsletter, town halls, change announcements, FAQs. 2028 Mutates Drafting goes to agents. Function moves into AI-rollout narrative, employee trust, backlash management. 2030 Continues, more strategic One of the most under-rated growth areas in HR-adjacent functions. AI rollouts that fail are usually comms failures, not technology failures. | |||
Change Management Supporting |
|||
2026 Continues Programme delivery, training plans, sponsor coalitions. 2028 Mutates Consolidates with internal comms and AI literacy. BCG's 10-20-70 framing applies: 70% of AI value is people and process, not technology. 2030 Continues, more strategic Owns the human-rate-of-change for the entire organisation. Permanent function, not a project capability. | |||
| Newly emerging | |||
AI Workforce Manager Newly emerging |
|||
2026 Does not yet exist as a role The work is being done by ad-hoc people in product ops, engineering and HR. 2028 New Microsoft 2025: 28% of managers considering this hire. Manages mixed teams of humans and agents, owns capacity planning across both. 2030 Continues One of the standard functions in any IT org of meaningful size. | |||
AI Trainer Newly emerging |
|||
2026 Does not yet exist as a role Training is fragmented across L&D, IT, and tool vendors. 2028 New Microsoft 2025: 32% of leaders considering hire within 12-18 months. Operationalises AI literacy across the workforce. 2030 Continues Permanent function, often inside a renamed L&D or Intelligence Resources department. | |||
Agent Orchestration Specialist Newly emerging |
|||
2026 Does not yet exist as a role Some platform engineers are doing this part-time, undefined. 2028 New Eightfold and Deloitte both name this as the most important hire of 2026. Designs and operates multi-agent systems across business processes. 2030 Continues Sits at the intersection of platform engineering, product and operations. Senior-leveraged, well paid. | |||
Context Engineer Newly emerging |
|||
2026 Does not yet exist as a discipline The term begins circulating in 2025 (Andrej Karpathy and others). 2028 New Cognizant publicly committed to hiring 1,000 Context Engineers in August 2025. Designs the data, prompts and retrieval scaffolding that makes models actually useful in a specific business. 2030 Continues Either a discrete discipline or absorbed into platform engineering. Either way the work is permanent. | |||
Intelligence Resources Lead Newly emerging |
|||
2026 Does not yet exist as a role The work is split across HR, IT, legal and digital ethics. 2028 New Microsoft 2025 framing: "Intelligence Resources departments will be essential for managing the interplay between humans and AI agents." Some companies still call this part HR or People & Culture, but the function is distinct. 2030 Continues Standalone function in most large enterprises. Owns workforce composition decisions across humans and agents. | |||