The acx report - Chapter 11

Frameworks & References.

The ACx Readiness Index, Watch List for 2026, our Responsible AI principles, and the full source material behind this report.

Access Full Report

* 01 Special Feature

The ACx Readiness Index.

A scored framework for assessing an organization's capacity to deliver on agentic customer experience. Seven dimensions, matched to the ACx Architecture Map and the three criteria of ACx success. Surveyed brands & enterprises. Scored 1–10.

2.9 /10

Average Enterprise

+ 7 /10

ACx Leaders

Strategy Definition

3.2 AVG

8.0 Leaders

Average org: AI investment linked to capabilities, not to a named customer outcome.

Leaders: every deployment tied to a defined outcome, baseline and accountable owner before work begins.

Data Foundation

4.0 AVG

8.1 Leaders

Average org: first-party data fragmented, customer identity not unified.

Leaders: unified data infrastructure with agentic access layers in production.

Delivery Acceleration

2.5 AVG

8.4 Leaders

Average org: measuring time-to-recommendation.

Leaders: 8-week standard to working systems, with production timelines built into engagement contracts.

Workforce & AI Proficiency

2.8 AVG

7.6 Leaders

Average org: high adoption, low mastery.

Leaders: 100% trained, role-specific credentialing, proficiency tied to recognition and advancement across all disciplines.

Industry Depth

2.4 AVG

7.9 Leaders

Average org: vertical context lost between strategy and production.

Leaders: deep domain expertise embedded in every delivery team, producing outputs that are commercially meaningful — not just technically correct.

Governance & Responsible AI

2.9 AVG

7.5 Leaders

Average org: governance treated as blocker or compliance exercise.

Leaders: ISO-aligned frameworks, policy as enabler, responsible AI as a procurement filter rather than a PR statement.

Measurement Rigor

2.2 AVG

7.3 Leaders

Average org: AI activity measured, outcomes not.

Leaders: 90-day ROI validation built into every deployment, with baselines set before launch.

Measurement Rigor scoring the lowest is the most telling finding in the Index. Most organizations do not actually know whether their AI investment is working. They know it is being used. They do not know what it is producing. That is the foundational problem the ACx standard is designed to correct — and the first conversation worth having with any organization serious about its readiness.


* 02 Special Feature

Designed for agents.

Today's customer experience is built for human eyes, thumbs and attention. Within ten years, the most valuable visitor to your brand may not be a person at all, it will be the agent acting on a person's behalf. The window to design for that future opens now.

Most CX surfaces — sites, apps, support flows, commerce funnels — are accidentally hostile to agents. They depend on visual layout, modal interruptions and brand storytelling that works for humans and confuses machines. The organizations that quietly build a parallel, agent-legible layer through 2027 will own the next channel before it is named.

Ten year horizon

2026

Phase 01

Lay the foundation

The next six months are about ACx fundamentals: structured content, semantic APIs, intent capture, agent-readable schemas. The plumbing nobody sees but every agent will depend on.

2027 → 2031

Phase 02

Two audiences in parallel

For roughly five years, every meaningful CX surface serves two consumers at once: a human who scans, taps and feels, and an agent that queries, negotiates and transacts. Brands that design for both win twice.

2032 → 2035

Phase 03

Agent-dominant

By the early 2030s, more transactions, comparisons and recommendations flow through agents than through human-driven sessions. The agent becomes the primary customer. Human visits become high-stakes exceptions.

2036 and beyond

Phase 04

The open question

Does the human channel disappear — the way most printed catalogues did — or does it persist as a deliberate, premium space the way browsers still persist alongside apps? We do not know. But the brands hedging both ways will outlast the ones that don't.

Built for humans

  1. Visual hierarchy, brand storytelling, motion and emotion

  2. Sequenced flows, persuasion arcs, considered choice architecture

  3. Trust earned through tone, craft and felt experience

Built for agents

  1. Structured data, machine-readable intent, clean semantic surfaces

  2. Programmatic negotiation, deterministic pricing and policy APIs

  3. Trust earned through reliability, provenance and verifiable claims

The brands that win the agent era are the ones that started building for it while it still looked optional.

The cost of waiting is not a missed launch, it is structural. Agent-readiness is not a layer you bolt on later. It is the architecture decisions you make now: how content is modelled, how identity is resolved, how transactions are exposed, how trust is verified. Brands treating 2026 as a foundation year — not a feature year — will be the ones with options when the channel mix tips.


* 03 Special Feature

Watch list 2026.

Six signals that did not make this year's main report, and will likely define next year's edition.

01

The Agentic Commerce Inflection

Shopping agents are moving from novelty to infrastructure. The first major brand to deploy a genuinely agentic commerce experience — one that remembers preferences, anticipates needs and transacts on behalf of the customer — will reset the conversion rate baseline for the entire Consumer arena. Watch Q3–Q4 2026.

02

AI Governance as Competitive Differentiator

The organizations building ethical AI frameworks and responsible AI credentials in 2026 are not just managing risk. They are building the trust infrastructure that enterprise procurement will require within 18 months. Responsible AI is becoming a procurement filter, not a PR statement.

03

The Fan Data Land Grab

First-party fan data is the most valuable and most underutilized asset in the Entertainment arena. The sports properties and media organizations building unified audience identity infrastructure in 2026 will have a 3–5 year head start on those that wait. The data is the moat. The agentic intelligence is what activates it.

04

Multimodal Agents in Consumer Experience

Text-based AI is the current enterprise baseline. The next wave — agents that reason across text, image, audio and video simultaneously — is already in early production deployment. CPG brands working on multimodal product experience today are twelve months ahead of where the Consumer arena will be in 2027.

05

The Services Firm Consolidation

The middle of the market is under pressure. Firms that are neither genuinely AI-native nor large enough to absorb R&D cost are being squeezed between tech-forward boutiques and offshore scale players. Consolidation through 2026–2027 is the likely outcome.

06

The Normalisation of Deviance in AI Deployment

The Challenger disaster was caused by the gradual acceptance of anomalies as normal. The same pattern is observable in enterprise AI today. Biased outputs accepted as "good enough." Safety flags dismissed because they have not caused a visible incident yet. Quality thresholds lowered incrementally. The organizations most at risk are the ones that have quietly redefined acceptable.


* 04 Special Feature

Using AI responsibly.

The ACx era will be defined not only by how fast organizations deploy agentic intelligence, but by how responsibly they do it. Speed without accountability is not a competitive advantage. It is a liability accumulating silently until it is not.

Apply is an AI-forward company. That means embracing AI as a core part of how we work and deliver value, and taking full responsibility for how we use it. The principles below are not a checklist. They are the values we hold ourselves to.

01

Human Judgment Remains Central

We are AI-forward, not AI-only. AI makes our people more capable. It does not replace their responsibility for the work. Every output — from code to copy and everything in between — is owned by the person who produced it. Our people are wholly accountable for what leaves our door, regardless of what tools were used to get there.

02

Transparency with Clients

AI is part of how we work across every department and every deliverable. All AI use follows our policies on confidentiality, data protection and responsible use without exception. We use AI platforms within our own environment to produce client work. We will never bring a tool, model or platform into a client environment that has not been approved for use within it.

03

Fairness and Bias Awareness

AI models can reflect the biases of their training data. That is a known limitation, and one our people are trained to catch. Across every discipline, human craft expertise is applied to all AI outputs before they reach a client. That is not optional, it is how we work. No AI output ships without a person accountable for it.

04

Regulatory Alignment

We only use AI models and platforms that align with the privacy and data regulations of the regions where we and our clients operate. Apply is ISO 27001 certified, with ISO 42001 pending. That certification is not a badge. It is the foundation our information security practices are built on, and it applies to everything we do, including how we use AI.

05

Right Model, Right Purpose

We choose AI models that fit the task, not the most powerful one available. Larger models consume more energy and compute. Using a frontier model where a lighter one works equally well is neither good practice nor responsible stewardship of shared resources. AI use at Apply should be deliberate.

06

Environmental Responsibility

AI has a real and growing environmental cost, not just in compute, but in the water used to cool the data centers that power these models. Our commitment to right-sizing model choice is one of the ways we act on that concern rather than just acknowledge it.

Responsible AI is not a constraint on what we build. It is the foundation that makes what we build worth trusting.

These principles evolve with the landscape. We review them annually — and sooner when significant changes in the technology, the regulation or the risk environment make it necessary. The organizations that treat responsible AI as a compliance exercise will be reactive. The ones that treat it as a competitive posture will be trusted — and in the ACx era, trust is the asset that compounds fastest.


* Sources & References

Further reading.

References cited in this report, with its source. All URLs active as of April 2026.

Chapter 01 & 04 · AI Project Failure & Scale

Gartner press release — Lack of AI-Ready Data Puts AI Projects at Risk (Feb 2025)

Through 2026, organizations are projected to abandon 60% of AI projects unsupported by AI-ready data

gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk

McKinsey — The State of AI (Nov 2025)

Approximately one-third of companies have begun scaling AI; nearly two-thirds have not yet begun scaling

mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

McKinsey Global Institute — Notes from the AI Frontier (Sept 2018)

Projected AI adding ~$13T to global GDP by 2030

mckinsey.com/featured-insights/artificial-intelligence/notes-from-the-frontier-modeling-the-impact-of-ai-on-the-world-economy

MIT NANDA Initiative — The GenAI Divide: State of AI in Business 2025 (July 2025)

95% of enterprise GenAI pilots fail to deliver measurable P&L impact

nanda.media.mit.edu/the-genai-divide-state-of-ai-in-business-2025

Chapter 01 · AI Adoption Acceleration

NBER Working Paper 32966 — The Rapid Adoption of Generative AI (Bick, Blandin & Deming, 2024)

Generative AI adoption running roughly 2× the speed of the internet at comparable ages (individual adoption)

nber.org/papers/w32966

Gartner press release — Hype Cycle for AI (Aug 2025)

AI agents and AI-ready data identified as the two fastest-advancing technologies on the 2025 Hype Cycle for AI

gartner.com/en/newsroom/press-releases/2025-08-05-gartner-hype-cycle-identifies-top-ai-innovations-in-2025

McKinsey Global Institute — The Economic Potential of Generative AI

Generative AI could add $2.6–4.4T annually across use cases; broader AI $13T by 2030

mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai

McKinsey — The State of AI

88% of enterprises now use AI in some form; only 23% report scaling agentic AI across the enterprise

mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

Chapter 02 & 03 · CX Quality & Ecosystem Design

Forrester — US Customer Experience Index 2025

CX quality has declined for four consecutive years; 25% of brands fell, only 7% improved

forrester.com/research/customer-experience-index

Forrester — Customer Journeys Connect Brand And Experience (2025)

Forrester’s ongoing argument that the linear funnel is dead; customer experiences span connected journeys, not single brands

forrester.com/blogs/customer-journeys-connect-brand-and-experience/

Chapter 07 · Google Cloud Platform

Google Cloud — Gemini Enterprise Agent Platform

Foundation model infrastructure, fine-tuning and deployment pipelines for enterprise AI

cloud.google.com/products/gemini-enterprise-agent-platform?hl=en

Google Cloud — BigQuery

Enterprise data warehouse and analytics foundation for agentic intelligence

cloud.google.com/bigquery

Google Cloud — Agent Builder

Orchestration layer for multi-step agentic workflows and conversational AI

cloud.google.com/products/agent-builder

Chapter 08 · Consumer Arena Market Data

eMarketer — Retail Media Advertising Forecast

Global retail media market growth; Walmart Connect, Amazon Ads and Target Roundel spend trends

emarketer.com/topics/topic/retail-media

Forrester — The State Of US Consumer Personalization, 2025

Consumer expectations for individualized brand experiences; retailer personalization gaps

forrester.com/report/the-state-of-us-consumer-personalization-2025/

Chapter 08 · Entertainment Arena Market Data

PwC Global Entertainment & Media Outlook

Streaming subscriber acquisition costs, retention rates and content economics

pwc.com/gx/en/industries/tmt/media/outlook.html

Deloitte — Digital Media Trends Survey

Audience attention fragmentation, streaming churn drivers and personalization demand

deloitte.com/us/en/insights/industry/technology-media-telecom/digital-media-trends-survey.html

Watch List · Normalisation of Deviance

Psychological Safety — Normalisation of Deviance

The Challenger disaster as a case study in how small accepted deviations accumulate into catastrophic risk

psychsafety.com/normalisation-of-deviance

NASA — Rogers Commission Report on Challenger

Original investigation into the organizational failures behind the Space Shuttle Challenger disaster

history.nasa.gov/rogersrep/genindex.htm

Responsible AI · Governance & Standards

ISO 27001 — Information Security Management

International standard for information security management systems

iso.org/standard/27001

ISO 42001 — AI Management Systems

International standard for artificial intelligence management systems and responsible AI governance

iso.org/standard/81230.html

NIST — AI Risk Management Framework

US federal framework for managing risk in AI systems across the full development lifecycle

nist.gov/artificial-intelligence/ai-risk-management-framework

Responsible AI · Environmental Impact

Goldman Sachs — AI Power Demand Report

AI data center energy and water consumption projections through 2030

goldmansachs.com/insights/articles/AI-poised-to-drive-165-increase-in-power-demand

Nature — Water Consumption of Large Language Models

Research on water usage in data centers powering large AI model training and inference

nature.com/articles/s41545-023-00274-6

Apply - Our Commitment

Build what comes next. Now.

The ACx era does not reward the cautious. It rewards the capable — the organizations with the data foundation, the talent depth, the partnership infrastructure and the accountability culture to move from ambition to production. That is the standard. That is the work.

 

We respect your privacy

We and selected third parties use cookies or similar technologies for technical purposes and, with your consent, for other purposes as specified in the cookie policy. Denying consent may make related features unavailable.