Designing IT Roles for an AI Era — A Talent-Strategy POV

Question (2026-06-02): As AI pushes humans toward higher-value work anchored in domain mastery and solution design, how should we structure IT roles inside an enterprise organised by job families? Filed alongside the citation set needed to take it to HR. Produced as a read-ahead note for the next IT LT as we define IT Talent strategy.

The thesis in one line

Rotate the organising axis of IT job families from technical discipline ("what technology do you operate?") to value contribution ("what irreducibly human value do you add to the outcome?"). Execution is becoming cheap; domain mastery, solution design, judgment, and accountability are becoming scarce. The structure should follow the scarcity.

The three-spine model (proposed)

Collapse current technical families onto three value "spines," with governance cutting across all of them. This maps directly onto Gartner's fusion team construct (organised by business/customer outcomes, shared accountability for business and technology outcomes).

Spine	Owns	Old families that fold in
Domain / Outcome	A business domain end-to-end; translates business intent into technical possibility (Gartner's "business technologist")	Business analysis, product, functional/app
Solution Design / Architecture	How problems get solved — patterns, build-vs-buy-vs-AI, non-functionals (security, cost, reliability, scale)	Enterprise/solution architecture, senior engineering
AI & Platform Orchestration	The leverage layer — agents, pipelines, harnesses, guardrails everyone builds on	Platform/SRE, DevOps, MLOps, data engineering
Trust & Governance (cross-cutting)	Risk, security, accountability for AI output — reframed from "produce-then-check" to assurance	Security, risk, QA

What thins out is the standalone execution tier inside each family — not because the work disappears, but because it's done by people wielding AI rather than a separate class of role. See Code Is Free for the premise (the bottleneck moves to defining, prioritising, instructing, accepting the work) and Bounded vs Unbounded Tasks for where agent autonomy is realistic.

The two things we cannot get wrong

1. Language and trust. We are shifting where value comes from, not bucketing people into high/low value. The anti-pattern is Standard Chartered's "lower-value human capital" gaffe (How Bosses Should Talk About AI (Economist)). The evidence: job insecurity impairs performance (Sverke 2019), but the damage is weaker when leaders are seen as honest/fair (Wang, Leicester — 16k employees). Trust is a performance modulator, not a soft nicety. Reinforced by FOBO (Fear of Becoming Obsolete).

2. The apprenticeship problem. If AI absorbs the junior rung where domain mastery used to be earned, where do future masters/architects come from? Design entry-level as AI-leveraged apprenticeship (juniors doing more under supervision, judgment reviewed by seniors) and protect deliberate inefficiency for mastery-building. Connects to the de-skilling risk in Cognitive Offloading and Will AI Make Us Dumber Method-Dependent Evidence. The org shape that operationalises this is the Hourglass Organization — pod at the top, lean middle, juniors learning the craft on the way up. Steven Brovich's named alternative to the "diamond" shape most companies are accidentally heading toward.

2026-06-27 — The talent archetype and the pipeline org shape now have names

Two updates land here from the same June 2026 ingest:

• The talent archetype the three-spine model implicitly hires for now has a name: the Expert Generalist (Martin Fowler / ThoughtWorks July 2025, popularised in A Leaders Guide to Advanced Team Structures (AWS Events)). Seven characteristics; curiosity, collaborativeness, customer focus, first-principles understanding are the four Brovich names. Hire for these, "not for the framework of the year — because that framework will change three times before this person's first performance review." Werner Vogels's Renaissance Developer is the same archetype from Amazon's angle: specialists broaden, generalists deepen, they meet in the middle.
• The org shape that protects the apprenticeship rung the page warns about is the Hourglass Organization. Brovich names the diamond (cut juniors, bulk middle with AI overseers) as the trap most companies are heading toward — "the middle hollows out; the top explodes." The hourglass keeps the pod (3–5 seniors + agents) at the top, but deliberately preserves and funds the junior apprenticeship rung. Matt Garman (CEO, AWS, 2025): "you absolutely want to keep hiring kids out of college." Empirical anchor: Anthropic Economic Index March 2026 — no systematic unemployment in exposed roles, but hiring of younger talent slowed ~14% since ChatGPT launched.

These updates don't replace the three-spine model — they sit underneath it. The three spines describe what value-axis IT job families should rotate onto; the Expert Generalist describes who staffs each spine's pod; the Hourglass describes how the org shape preserves the rung that produces tomorrow's expert generalists. For IT-LT communication: lead with the three-spine model (the what), defend with the Expert Generalist (the who), make the Hourglass commitment visible (the how we'll still have seniors in 2034).

The EI Survey June 2026 also adds a non-obvious empirical air: people who use Claude in the most automated way are most optimistic about pay, job security, and meaning. This doesn't dismiss FOBO — it's a usage-cluster effect, not a population effect — but it's the cleanest empirical counter the vault has to the dominant fatalistic-headline read. Useful when defending the three-spine + hourglass commitment against "but won't AI just take everyone's job anyway" pushback.

Citation map (for HR)

Tier 1 — institutional research / peer-reviewed (lead with these):

• Gartner — "All IT Work Will Involve AI by 2030" (700+ CIO survey, Jul 2025): by 2030, 0% of IT work done by humans without AI, 75% augmented, 25% AI-alone. → the structural premise.
• WEF — Future of Jobs Report 2025 (1,000+ employers, 14M workers): 85% adopting AI to augment not replace; net +78M jobs by 2030; 39% of core skills change. → augmentation framing + reskilling mandate.
• Deloitte — The Skills-Based Organization (1,021 workers / 225 execs): 63% of work already falls outside core job descriptions; 81% crosses functional boundaries; outcome-organised firms 63% more likely to achieve results, 79% more likely to deliver positive workforce experience. → evidence for "organise around value, not the rigid job."
• Gartner — Fusion Teams: multidisciplinary teams blending tech + business-domain expertise, organised by outcomes, shared accountability. → the three-spine model's external anchor.
• Gartner HR — "AI Will Create More Jobs Than It Eliminates Beginning 2028" (May 2026): "jobs chaos" — reconfigure/redesign/splinter/fuse tens of millions of roles a year. → urgency to design the axis deliberately now.
• Gartner — "85% of Service Leaders Expanding Human Responsibilities" (Apr 2026): redesign, not reduction, even in the most-exposed function. → counters the layoff read.
• Sverke et al. (2019) + Wang (Leicester), via How Bosses Should Talk About AI (Economist). → the trust/performance guardrail (peer-reviewed).

Tier 2 — practitioner / governance (support, don't lead):

• Code Is Free (Lopopolo; triangulated with Cherny, Karpathy) — value relocates from execution to judgment.
• Learning Software Engineering During the Era of AI (Raymond Fu, TEDxCSTU) — "AI raises the floor, engineers raise the ceiling." Caveat: small TEDx, uncited stats — framing only.
• Governing AI Agents at Scale (Glean + Cvent, CXOTalk) / AWARE Framework / Five AI Risks That Can Get You Fired (IBM Technology) — the Trust & Governance spine.

Not used (flagged so we don't over-claim): the "T-shaped → π-shaped" career model — unsupported in-vault and partly contradicted by Fu's full-stack argument; dropped in favour of the Deloitte skills-based framing. Several Gartner items are press releases summarising paywalled reports — fine for a discussion note; pull full research notes if this graduates to a board-level strategy.

External reference links

1. Gartner — All IT Work Will Involve AI by 2030: https://www.gartner.com/en/newsroom/press-releases/2025-10-20-gartner-survey-finds-all-it-work-will-involve-ai-by-2030-organizations-must-navigate-ai-readiness-and-human-readiness-to-find-capture-and-sustain-value
2. WEF — Future of Jobs Report 2025: https://www.weforum.org/publications/the-future-of-jobs-report-2025/
3. Deloitte — The Skills-Based Organization: https://www.deloitte.com/us/en/insights/topics/talent/organizational-skill-based-hiring.html
4. Gartner — Fusion Teams: https://www.gartner.com/en/articles/fusion-teams
5. Gartner HR — AI Will Create More Jobs Than It Eliminates Beginning 2028: https://www.gartner.com/en/newsroom/press-releases/2026-05-13-gartner-hr-research-reveals-ai-will-create-more-jobs-than-it-eliminates-beginning-in-2028
6. The Economist — How should bosses talk about AI?: https://www.economist.com/business/2026/05/28/how-should-bosses-talk-about-ai
7. Gartner — 85% of Service & Support Leaders Expanding Human Responsibilities: https://www.gartner.com/en/newsroom/press-releases/2026-04-28-gartner-survey-finds-eighty-five-percent-of-service-and-support-leaders-are-expanding-human-agent-responsibilities-despite-expectations-of-mass-ai-layoffs

What I'd ask of the IT LT

1. Agree the organising principle — value/skills contribution over technical discipline (Deloitte, Gartner).
2. Decide whether the three-spine / fusion-team model is the right frame to re-architect our IT job families against.
3. Commit to the trust-narrative guardrail before communicating any restructuring.
4. Name an owner for the apprenticeship-pipeline question — longest lag, largest downside.

2026-06-13 — "Don't hire prompt engineers; redeploy your platform/SRE/security engineers"

The 7 Skills You Need to Build AI Agents (IBM Technology) (Bri Kopecki) supplies a concrete competency model that sharpens the talent-strategy conclusion of this note. Her seven agent-building skills — system design, tool/contract design, retrieval, reliability, security, eval & observability, product thinking — are, by her own argument, mostly re-pointed backend / distributed-systems / security disciplines, not net-new ones. Five of the seven map straight onto competencies our platform/SRE/security engineers already hold.

The implication for talent strategy: the AI-talent question is not "where do we hire prompt engineers" but "how do we redeploy and upskill the platform, SRE, and security engineers we already have." This is a practitioner-side corroboration of the Skill Change Index (SCI) thesis — most "agent" skills score as re-application of existing expertise (low skill-change), not greenfield reskilling — and it gives the AI & Platform Orchestration spine above a named skill inventory to build job descriptions and skill corridors against. It also names the role transition cleanly as a move into Agentic Engineering (Kopecki's "agent engineer"), rather than a new job family invented from scratch. Caveat: single channel/author, practitioner explainer — use as a competency-model scaffold, not as institutional evidence on the scale of the Gartner/Deloitte Tier-1 anchors.

2026-06-27 — Central-bank Tier-1 anchor for the apprenticeship problem

Does AI Adoption Improve Productivity (BOK Issue Note 2026-12) elevates "the apprenticeship problem" from a precautionary footnote in this note to its own labelled policy section in a Bank of Korea Issue Note. The Suh/Oh/Yoon line:

"Traditionally, standardized tasks have functioned as the learning pathway through which junior and early-career workers accumulate domain knowledge and work intuition... If standardized tasks are replaced by AI, this traditional skill formation pathway may be weakened. In the short term, productivity may improve, but in the long term there is a risk that the supply base of human resources capable of performing open tasks may itself be undermined."

Their three operational countermeasures are directly usable as the apprenticeship-design line item in a 3-spine job-family proposal: (1) observe→assist→lead structure putting juniors into open tasks at an early stage rather than gating them behind standardized work; (2) senior time saved by AI reinvested in mentoring / coaching / pair work, not just into more senior output; (3) for AI-automated tasks, deliberate learning structures so verification & exception-handling skills are accumulated from the junior stage. Pair with their Standardized vs Open Tasks framework when distinguishing which job-family rungs are exposed to the absorption risk.

This raises this note's confidence on the apprenticeship-problem axis specifically — it now has a representative national household survey + central-bank policy framework behind it, not just precautionary commentary. Worth adding the BOK paper to the Tier-1 citation map the next time this gets prepared for HR.

Cross-links

• Applied POVs (same org context) · Elevating Manila IT — A 10X-but-not-Hustle Point of View · DRAG for AI Upskilling at Manila IT Site
• Concepts · Code Is Free · Bounded vs Unbounded Tasks · Harness (LLM Agents) · FOBO (Fear of Becoming Obsolete) · Cognitive Offloading · AWARE Framework · Human in the Loop
• Sources · How Bosses Should Talk About AI (Economist) · Learning Software Engineering During the Era of AI (Raymond Fu, TEDxCSTU) · Governing AI Agents at Scale (Glean + Cvent, CXOTalk) · Five AI Risks That Can Get You Fired (IBM Technology) · GCC Philippines Summit 2026 (PHx)
• GCC talent data (fresh citations) · Skill Change Index (SCI) (~6,800 skills; routine→judgment repricing) · Frontier GCC (10-30-50 leadership benchmark) · Headcount-to-Value Pivot · Chief GBS Orchestrator

Brand fodder

Internal-first piece (read-ahead note to HR VP + IT VP-Talent for IT LT). A public LinkedIn/Medium version writes from the same spine: "Stop organising IT around who operates the technology. Organise around who owns the outcome, who designs the solution, and who builds the leverage — with AI as the universal execution layer underneath." On-thesis for the senior-IT-leader brand; lead with the Gartner 0%/75%/25% stat and the Deloitte "63% of work already falls outside the job description" hook.