Mastering AI-ready data: five steps to an effective consent flywheel

Picture a global retailer trying to keep shelves stocked during sudden supply chain shocks, while also tailoring promotions to customers who expect instant, personalised offers. Or a bank balancing stricter regulations with the pressure to provide smooth digital experiences. Leadership knows that many of the clues to meeting these challenges lie in their data — but too often that data feels fragmented, slow to move, and difficult to trust.

What if instead, permissioned data with usage consent was always in motion, continuously refined and redistributed, fueling every decision and every AI interaction with confidence? That is the promise of an AI-ready data (AIRD) consent flywheel — a self-reinforcing concept for approaching data strategy for AI systems that helps organisations continuously collect, manage and ethically use customer data with consent.

How can this approach help to make your data AI-ready? These five steps are key:

1. Harness the flywheel effect to set your data in motion

Too often, organisations treat data like a static library, which is accessible only if you know where to look, and rarely updated. The flywheel changes that. Instead of being stored and forgotten, data is continuously refined and enriched, with permission controls applied and redistributed — a living system that improves with every turn of the wheel.

At the heart of this flywheel is AI-ready data. Without AIRD, the flywheel cannot generate momentum; with it, every cycle produces more value and greater confidence in AI.

AIRD is discoverable, trusted and accessible (i.e., with traceable usage consent) to both humans and AI agents. It’s also high-quality, rich in business context and properly formatted, with usage permissions well-governed. Traceable, governed data consent is critical to making data accessible because it confirms legal use conditions, protects privacy and allows sharing without risk.

But how can you use it? Integrating an AIRD consent flywheel into your own company is less about buying a product and more about embedding the principles into your data and AI stack.

At the EY organisation, we accomplished it by building on three interlocking components — components that apply regardless of the type and scale of your organisation:

• A data product marketplace where trusted, contextual data assets are produced and maintained at scale for teams and clients
• Optimisation and model training, helping ensure AI learns, adapts and stays aligned with business context over time
• An agentic platform, the engine that allows AI agents to analyse AIRD and apply it directly to business outcomes

When these parts work together, data shifts from a background utility to a front-line enabler of innovation.

2. Permission your dataflows to speed up innovation

AI innovation is moving at breakneck speed. We see benchmarks shattered in large language models (LLMs), billions invested in AI infrastructure, and falling costs of compute power. Yet within many enterprises, the movement of data remains painfully slow, with approvals bottlenecked, processes outdated and valuable data left unused.

A data green lane can help. A data green lane is a pre-approved, policy-enforced route that allows trusted data to move quickly into AI and business processes while the right guardrails are applied by default. Rather than having every data request stop at a red light, green lanes accelerate progress while still enforcing the right guardrails.

At EY, we use a data product marketplace to bring this concept to life, making AIRD instantly available to teams and clients who need it.

3. Deploy data agents

Enterprises often face a widening gap: AI capabilities evolve at lightning speed, but data estates — the infrastructures that help companies systematically manage all of their corporate data — struggle to keep up. Data agents help close that gap. Think of them as smart helpers that work in the background to keep your data estate AI-ready.

These agents can:

• Onboard and validate new data sources.
• Enrich metadata and business context.
• Build and adjust data structures on demand.
• Help ensure governance, trust and oversight remain intact.

At EY, data agents on the agentic platform are supercharging the creation of our AIRD estate, helping ensure our data keeps pace with shifting business needs.

4. Leverage synthetic data

Even the best AI is only as good as the data it learns from. Bias, incompleteness or scarcity in datasets can cripple performance.

Synthetic data is a powerful tool to help. By creating realistic, representative data where gaps exist, organisations can:

• Reduce bias and facilitate fairness.
• Expand training datasets where real-world examples are limited.
• Safely test sensitive scenarios without exposing actual data.

Our early client work highlights the possibilities:

• At one manufacturing organisation, synthetic data for predictive maintenance pointed to opportunities worth around US$11m in potential annual savings.
• In a financial services firm, synthetic records demonstrated the potential to cut onboarding failures by up to 70%.

At EY, we use a simulation hub to experiment with synthetic datasets under strict guardrails, balancing speed and responsibility.

5. Train talented people to reimagine AIRD-led ways of working

Technology alone can’t deliver on the promise of agentic AI. The shift to AIRD demands a new class of professionals, beyond traditional data scientists or analysts. These individuals are builders and curators of continuously evolving data estates, working hand-in-hand with AI agents.

Other roles will expand too. According to a recent EY survey, 87% of tax leaders already view generative AI (GenAI) as transformative for their teams – part of a wider trend where the World Economic Forum (WEF) predicts two-fifths (39%) of workers’ skill sets will be transformed by 2030. At the EY organisation, we believe that the future of work will be defined by human-AI collaboration, where AIRD helps ensure both sides speak the same language.

Organisations that invest in developing this talent will find themselves not only keeping pace with AI, but setting the pace.

Building AI-ready data foundations for agentic AI

With the AIRD consent flywheel, organisations can unlock a virtuous cycle of accelerating returns – better outcomes at lower cost – that cascades through their clients, and theirs in turn.

Agentic AI is not just another incremental upgrade. It is a step-change in how work is done, decisions are made, and value is created. To seize this opportunity, organisations must stop treating data as a passive resource and start treating it as the active fuel of transformation.

Shifting your data estate to AI-ready data is the single highest ROI activity you can undertake to prepare for this new era. With an AIRD consent flywheel in motion, businesses can unlock the full promise of agentic AI. The organisations that master this will drive growth, resilience and trust, and do it confidently, responsibly and at scale.