Clektive Thinking: What Snowflake’s latest innovations mean for retail leaders
AI has dominated headlines for the last two years, but the conversation is already moving on. The next evolution is agentic AI: intelligent agents that can reason, make decisions and act on behalf of people. For retailers, this represents one of the biggest shifts since the rise of ecommerce.
In a recent episode of Clektive Thinking, Clekt CEO Andy Tudor sat down with Paul Winsor, Head of Retail EMEA at Snowflake, to unpack the latest innovations Snowflake are bringing to the market, what they mean for retailers and, more importantly, the practical steps businesses should be prioritising now to prepare for the agentic AI era.
The key themes shaping retail in the agentic AI era.
One big misconception surrounding agentic AI is that organisations should start by looking at the technology. According to Paul, that’s the wrong place to begin.
“Where we’re seeing conversations now with retailers and consumer brands is about the whole agentic era that we’re entering into. The excitement around that in terms of what the possibilities are, but also making sure that it’s set up correctly, that we don’t create more inconsistencies. I think really the best place to start is not to talk about the capability but talk about the priorities of the retailers.”
Those priorities fall into three clear themes.
1. Attracting increasingly conscious consumers
Today’s consumers expect more than competitive pricing. They want value, relevance and experiences that feel personalised.
As Paul explains:
“How do you attract more conscious consumers? These are consumers today that I’m sure we’re all feeling the same about, which is we’re seeking value and we’re seeking good products. Retailers are constantly looking at how they attract more of those conscious consumers. The Agentic capability is going to supercharge and accelerate the way that we can do that.”
Instead of relying on historical reporting, AI agents can continuously analyse customer behaviour, market trends and external signals to recommend the next best action, whether that’s a targeted promotion, a personalised recommendation or a more relevant customer experience.
2. Building more resilient operations
Supply chains have become increasingly unpredictable over the last few years. Economic uncertainty, geopolitical disruption and changing customer demand mean retailers need to react much faster than traditional reporting allows.
As Paul puts it:
“Supply chains weren’t built for this level of disruption… Building resilience into your operation, to make sure that you can be aware of situations impacting your operations, is really important. We’re seeing agentic strategies being built into those resilient operations.”
Rather than replacing people, agentic AI gives operational teams earlier visibility of risks and helps them make faster, better-informed decisions.
3. Preparing for the rise of agentic commerce
Perhaps the biggest shift retailers should be thinking about isn’t happening inside their business at all. It’s happening with consumers. Paul describes a future where shoppers increasingly rely on AI agents to research products, compare prices and even complete purchases on their behalf.
“Consumers will eventually start using agents that they want to use to help them find products and services. Those agents are going to have to go and seek those products and services from the wider internet rather than just directly with retailers.”
That fundamentally changes how brands are discovered, compared and chosen. Retailers that prepare their data today will be far better positioned when agentic commerce becomes mainstream.
What does it mean to become agentic AI ready?
The answer isn’t buying another AI tool. It’s ensuring your data is ready for AI to understand. Paul explains that many retailers have already invested in modernising their data estate. The next challenge is making that data usable by AI agents.
“Retail customers have actually invested in modernising their data estate. We’re having that more mature conversation around, ‘I’ve got my data in a place that has good quality, that the business recognises as being the truth of the business.’ But to introduce agents to work alongside knowledge workers, you can’t just introduce agents on top of your data without building this semantic layer that connects the agent to the right set of data so it can answer the question with context.”
That semantic layer becomes the bridge between business language and technical data. Instead of simply retrieving information, AI agents understand what the business is trying to achieve, giving people not just answers, but answers with context.
From answers to action
One of the strongest examples discussed was the role of a retail category manager. Historically, identifying pricing changes across thousands of products could take hours or even days. With agentic AI, that process becomes almost immediate.
As Paul explains:
“A category manager is there to grow volume of sales, have the right products in the right assortment and competition is fierce when it comes to price. How can an agent accelerate the answer and the context of that answer if a competitor is selling a product cheaper than the retailer today?”
Rather than manually checking competitor websites, AI agents can monitor pricing continuously, compare internal and external data, surface opportunities and even recommend the next action. The result is a dramatic reduction in the time between identifying an issue and responding to it.
“The time to insight and the time to action is reducing now so fast.”
The consequences of getting it wrong
Introducing AI across an organisation without governance creates a different problem. Instead of helping departments work together, AI can unintentionally reinforce silos. Paul illustrates this with a simple but powerful example.
“You might have a knowledge worker and an agent figuring out how to reduce inventory to reduce costs in your business. At the same time you might have another knowledge worker trying to grow sales around products and thinking about putting a product on promotion. The agent that’s going to suggest putting something on promotion might choose the same product another agent has just decided to reduce stock on.”
Both agents have completed the task they were given. Neither understands the wider business objective.
What is an agentic control plane?
To avoid these conflicts, Snowflake introduces the concept of an agentic control plane. Think of it as air traffic control for AI agents. Rather than allowing every AI agent to operate independently, the control plane provides shared business context, governance and visibility across the organisation. It ensures every agent understands not only its own objective but also the wider business priorities. As Andy puts it:
“It’s about providing the air traffic control to the agents.”
This coordinated approach allows marketing, finance, operations and supply chain to work towards the same goals rather than competing with one another.
A practical use case: Social listening powered by agentic AI
One of the most interesting examples shared by Andy was of a fashion retailer using agentic AI to monitor how products were being discussed across social media, to improve customer conversations.
“This particular fashion retailer was exploring different social media channels, understanding how their brand was being talked about in colloquial language, bringing that data back into their organisation, enriching their product descriptions with that social listening data, and then pushing that data into the hands of store associates so they could converse with customers in the same language.”
Instead of relying on periodic customer research, the retailer was continuously learning how customers naturally described products and immediately feeding those insights back into both digital and in store experiences. Paul highlights why this matters.
“The time to insight and action… we’re getting down into some incredible near real time decision making.”
Are retailers ready for what’s coming next?
Technology is advancing so quickly that traditional strategic planning is becoming almost impossible. When Andy asked what retail might look like in five years, Paul’s answer perfectly captured the pace of change.
“Somebody asked me yesterday what I thought the vision was for retailers in the next five years. I said, ‘Can we reduce that to five months?’… The idea that we used to say, ‘Where do you want to be in five years?’ Forget it.”
The challenge isn’t understanding what agentic AI is. It’s operationalising it across the business. Retailers now need to think beyond individual AI tools and towards connected data, governance, enterprise-wide adoption and measurable business outcomes.
Five practical steps to become agentic AI ready
Paul offered some straightforward advice for organisations that haven’t yet started.
- Start with business outcomes
Don’t begin with AI. Start with the problem you’re trying to solve. Is it pricing? Customer experience? Inventory optimisation? Campaign performance? Think about it by function, by persona and by outcome.
- Identify where AI can support knowledge workers
Look for repetitive, data intensive tasks where employees spend time gathering information before making decisions. These are often the quickest opportunities for agentic AI to deliver value.
- Build an AI ready data foundation
Reliable, connected and contextual data remains the single biggest success factor. Without it, even the most advanced AI models will struggle.
- Measure your digital workforce
One fascinating example shared in the podcast described organisations already measuring the performance of AI agents alongside human employees. How effectively is this agent performing for this knowledge worker? Is it contributing to business outcomes?
- Experiment early
One of Paul’s strongest recommendations is don’t wait. Retailers shouldn’t delay adoption until the technology feels perfect.
“I don’t think this is a capability that anybody should be sitting to the side and waiting for. It has too much return on investment… This is not about wait and see. It’s about getting yourself prepared.”
Start with internal use cases, improve accuracy over time and allow both people and AI agents to learn together.
Final thoughts
Agentic AI isn’t simply another technology trend. It represents a fundamental shift in how retailers will operate, make decisions and engage with customers. The businesses that succeed won’t necessarily be those adopting AI first. They’ll be the ones investing in trusted data, strong governance and clear business outcomes, giving AI the context it needs to become a genuine extension of their workforce rather than another disconnected tool.
As Paul concludes, now isn’t the time to sit back and wait. The organisations building those foundations today will be the ones best placed to lead tomorrow.
Want to discuss your agentic AI strategy?
At Clekt, we help retailers build the data foundations that enable AI to deliver meaningful business outcomes. Whether you’re modernising your data estate, designing a semantic layer or exploring your first agentic AI use case, our team can help you take the next step with confidence.
