AI has changed a lot of things in the user experience industry — the way we design, the tools we use, the outcomes we deliver. We need new structures and new operating models to sustain ourselves in the era of AI. I've been watching a number of thoughtful leaders talk about the changes they're actively deploying so their teams are prepared for what's next. That has inspired me.
What I've been focusing on is this: which core competencies do design and research teams need most in the era of AI, and what should the future competency matrix look like?
Some Interesting Statistics
A few numbers from the 2026 AI in Design survey:
- 73%of designers report that expectations of them are rising — faster turnaround, higher output volume, broader scope.
- 28%of design leaders have made any formal organizational change. Only 13% have updated performance review criteria or hiring practices. Half report no structural changes to their teams at all.
- 65%of designers say they're doing work that traditionally belonged to PM or engineering — implementation, prototyping, validation. At the same time, 40% say their PMs and engineers are doing more design work.
- 24% → 70%is the jump in peer learning as a primary source in a single year, while learning from leadership recommendations dropped from 32% to 16%.
My guess: the more forward-looking design teams have already begun adjusting their operating models. Through the rest of 2026, more teams will follow — defining new roles, updating performance criteria, and changing team composition.
Four (Possible) Maturity Levels
If we divide the growth path of future UX practitioners into four levels, what I'm seeing is that these levels are defined by their impact on Systems, Evaluation, Building, and Governance.
From there, we can extract the core dimensions that move a practitioner from a lower level to a higher one. Those are:
Experience Orchestration & Building
From designing static screens to directly orchestrating code and agents.
UX Evaluation & Quality Assurance
From running usability tests to automated quality checks before launch.
Insight, Judgment & Business Advancement
From conducting research to influencing business and product decisions.
Knowledge Base & Governance
From a static knowledge base to building and governing an enterprise-level context brain.
Let's break down each dimension.
The Matrix
Dimension 1 · Experience Orchestration & Building
From designing interface to directly orchestrating code and agents
| Level | Behaviors & Deliverables | Assessment Criteria |
|---|---|---|
L1Novice Operator |
Uses ChatGPT/Claude to help write UI copy, generate design direction, or summarize research notes | Tool use is habitual rather than occasional; individual output efficiency lifts measurably |
L2Augmented Builder |
Builds interactive front-end prototypes directly in Cursor / v0 / Bolt rather than static handoff files | Deliverables shift from Figma files to runnable prototypes; cross-platform validation gets faster |
L3Experience Architect |
Designs and orchestrates lightweight agent workflows; builds synthetic users to pre-test design directions | Can stand up an agent prototype with memory and tool use, via code or a no-code platform |
L4System & Governance Leader |
Defines generative UI specifications; works with architects to establish org-wide protocols for when AI should emit which design system component | Standards exist and are enforced; generated output is controllable and compliant |
Dimension 2 · UX Evaluation & Quality Assurance
From running usability tests to automated quality checks before launch
| Level | Behaviors & Deliverables | Assessment Criteria |
|---|---|---|
L1Novice Operator |
Identifies visual errors, misalignment, and logic gaps in generative UI through routine expert review | Reliably flags non-compliant elements in AI-generated interfaces |
L2Augmented Builder |
Breaks vague UX principles into a structured rubric — tone, accessibility, cognitive load, layout priority | The rubric is unambiguous enough for non-designers or machines to apply consistently |
L3Experience Architect |
Builds LLM-as-a-Judge evaluation systems; writes eval prompts, configures test scripts, tracks UI pass rates | Automated test suites catch the large majority of generation-quality issues before launch |
L4System & Governance Leader |
Establishes closed-loop offline/online evaluation; wires Evals into CI/CD and calibrates the judge against real post-launch user behavior data | Experience pass rate does not degrade through model upgrades or code refactors |
Dimension 3 · Insight, Judgment & Business Advancement
From conducting research to influencing business and product decisions at a system level
| Level | Behaviors & Deliverables | Assessment Criteria |
|---|---|---|
L1Novice Operator |
Uses AI to quickly organize interview transcripts, extract high-frequency pain points, and produce summary reports | Time to basic insight drops substantially |
L2Augmented Builder |
Distills system-ready insight — structured output that can feed a prompt or knowledge base directly | Research output is machine-consumable, not only human-readable |
L3Experience Architect |
Exercises judgment in choosing research questions: skips what AI can cheaply validate and focuses on high-risk, high-value uncertainty | Findings directly trigger a change in direction or a stop-loss decision, rather than a confirmation |
L4System & Governance Leader |
Translates insight across functions and pre-aligns stakeholders ahead of readouts, in the language of PM, Legal/Compliance, and the CEO | Drives company-wide strategic decisions, not changes to a single page |
Dimension 4 · Knowledge Base & Governance
From building knowledge base to building and governing an enterprise-level context brain
| Level | Behaviors & Deliverables | Assessment Criteria |
|---|---|---|
L1Novice Operator |
Organizes personal research documents and design specs so they're easy to retrieve individually | Personal documentation is well structured and findable |
L2Augmented Builder |
Proactively uploads research assets to a team-level knowledge base for others to query | Contribution is real and tracked, not aspirational |
L3Experience Architect |
Helps build a synthetic-persona knowledge base, ensuring it is unbiased and grounded in verified truth | Synthetic users achieve high agreement with real user feedback in blind parallel tests |
L4System & Governance Leader |
Establishes data governance standards for research assets — preventing anyone from making a rash product decision off a single unverified AI summary | Data-use compliance policy is set; anti-misuse and anti-hallucination guardrails sit at the knowledge-base infrastructure level |