Generative AI’s real bottleneck isn’t technology. It’s design.

Over the past two years, generative AI has moved from curiosity to corporate priority. Large language models can draft marketing campaigns, analyse documents, write code and answer complex questions in seconds. Yet despite the hype, most organisations are still struggling to move beyond experimentation.

The reason might surprise you.

The biggest barrier to scaling generative AI isn’t the technology itself. It is the way organisations design experiences around it.

This challenge is especially visible in retail. Retailers have no shortage of potential AI use cases, from personalised marketing and merchandising insights to supply chain optimisation and store operations support. Yet many initiatives stall after early pilots because the tools never become part of everyday workflows.

The Pilot Trap

Industry research shows that generative AI adoption is widespread but shallow. Around 65% of organisations report regularly using generative AI, nearly double the rate from just a year earlier. However, only a small percentage have successfully embedded the technology into core business processes.

Similarly, Gartner estimates that fewer than 30% of generative AI initiatives move beyond the pilot phase, and even fewer deliver measurable financial impact.

In other words, companies are experimenting, but very few are transforming.

Retail is a clear example. Many retailers are testing AI tools to generate product descriptions, analyse customer reviews or support marketing teams. But far fewer have integrated AI deeply into areas like assortment planning, demand forecasting or real time customer engagement.

The issue is rarely a lack of ideas. Most organisations actually have too many.

The real problem is knowing where to start, what to prioritise and how to move from isolated experiments to a coordinated roadmap.

We’re Using AI Through Outdated Interfaces

Think about how most people interact with generative AI today. Whether it is an internal corporate tool or a public platform, the interface usually looks the same: a simple chat box or search bar.

The model may be capable of complex reasoning and creativity, but the experience around it remains primitive.

A chat interface is excellent for asking a quick question. It is far less effective for navigating multi step work processes like designing a marketing campaign, managing a sales pipeline or planning a product launch.

In retail environments, the limitations become even clearer. A merchandiser trying to plan seasonal assortments, a marketer optimising promotions or a store manager analysing performance needs tools that connect data, workflows and decisions.

A chat box alone rarely delivers that.

As a result, users often swing between two extremes

• Over-trusting the output and assuming the AI is always correct
• Abandoning the tool entirely when it produces weak or inconsistent results

Speed and fluency can impress users, but they do not automatically create understanding or trust.

The real challenge is not building smarter models. It is designing better AI experiences and aligning them to real business processes.

Design Is a Strategic Capability

More than a decade ago, research showed that design is not just about aesthetics. Companies that invest heavily in design outperform industry benchmarks in both revenue growth and shareholder returns.

That same principle now applies to AI.

Organisations that succeed with generative AI will be those that rethink how people interact with intelligence, creating tools that integrate naturally into the way employees actually work.

For retailers, that might mean AI systems that support store associates in real time, help planners explore assortment scenarios or allow marketing teams to test campaign ideas rapidly across channels.

But before building anything, organisations must answer a more fundamental question:

Where will AI create the most value for our business?

That requires a structured approach, identifying opportunities, evaluating impact and building a clear roadmap.

Four Design Principles for AI Native Systems

Emerging best practices suggest that successful AI systems share four key characteristics: clarity, continuity, depth and collaboration.

1. Clarity. Make AI Reasoning Visible

One of the biggest barriers to AI adoption is opacity.

When users cannot see how an AI system arrived at an answer, trust erodes quickly. People either accept outputs blindly or dismiss them entirely.

AI systems must make reasoning more legible by explaining how conclusions were reached, showing relevant data sources and highlighting uncertainty or confidence levels.

For retailers, this might mean showing the drivers behind demand forecasts, explaining pricing recommendations or revealing the data behind product suggestions.

Transparency allows users to challenge, refine and ultimately trust the system.

2. Continuity. AI Should Remember the Work

Most AI tools treat every interaction as a fresh start. But real work does not happen that way.

Projects evolve. Conversations build on previous insights. Teams accumulate knowledge over time.

AI systems that support real productivity need organisational memory, the ability to understand previous interactions, track progress and anticipate what comes next.

In retail, this could allow AI tools to learn from past promotions, track the performance of product categories over time or build institutional knowledge about customer behaviour and seasonality.

3. Depth. Move Beyond Answers to Workflows

Today’s generative AI tools are often optimised to deliver answers. But business value rarely comes from answers alone.

Real impact comes from automating and orchestrating entire workflows.

Imagine a retail AI system that does not just suggest marketing copy but also pulls customer insights from loyalty and transaction data, generates campaign variants for different segments, tests messaging across channels and recommends optimisation strategies based on performance.

That level of depth transforms AI from a novelty into an operational engine.

4. Collaboration. Human and AI Working Together

The phrase human in the loop often implies that people exist mainly to correct AI mistakes.

But the most powerful AI systems enable something more dynamic: collaborative intelligence.

In these environments, humans and AI continuously interact, questioning assumptions, refining ideas and iterating on outputs.

The goal is not to replace human judgment. It is to amplify it.

From AI Experiments to an AI Roadmap

Many organisations today are in the same position. They have experimented with AI tools. They have seen promising results in small pilots. But they lack a clear path to scale.

The next step is not simply deploying more AI tools.

It is identifying the right opportunities and building a structured roadmap that connects AI initiatives to measurable business outcomes.

For retailers, that could mean prioritising use cases such as personalised marketing, demand forecasting, merchandising optimisation, customer service automation or store operations intelligence.

The key is to approach AI strategically rather than opportunistically.

Turning AI Opportunity Into Action

Generative AI has already proven its technical potential.

The organisations that will benefit most are those that take a structured approach to identifying opportunities, designing meaningful AI experiences and building a practical roadmap for adoption.

For retailers exploring how AI could transform their business, the first step is understanding where the real opportunities lie.

To explore where generative AI can create value in your organisation and how to build a clear roadmap for adoption, connect with our AbsoluteLabs AI experts and discover how to turn experimentation into real business impact.

Sources

McKinsey & Company — The State of AI in 2024: Generative AI’s Breakout Year
McKinsey & Company — The Economic Potential of Generative AI: The Next Productivity Frontier
Gartner — Gartner Poll Finds 55% of Organisations Are in Piloting or Production Mode with Generative AI
Gartner — Gartner Survey Finds Generative AI Is Now the Most Frequently Deployed AI Solution in Organisations
KPMG — From Pilots to Production: Scaling AI in the Enterprise
McKinsey & Company — The Business Value of Design

‍