POLYWOOD is North America's largest direct-to-consumer outdoor furniture brand, with more than 150,000 product variations manufactured in-house.
After years on a heavily customized Magento build that consumed the majority of its engineering capacity, the company migrated to Shopify — and immediately gained something it hadn't had in years: time to experiment.
With engineering resources no longer tied to platform maintenance, POLYWOOD began embedding AI across virtually every function — from development and customer service to manufacturing, product discovery, and preparation for AI-driven shopping channels.
POLYWOOD's AI transformation:
- Virtually all development resources redirected from platform maintenance to value creation and AI innovation
- AI embedded across development, creative, customer service, and operations
- 100% AI-assisted coding across the Shopify development team
- Conversational AI product discovery deployed across a 150,000-SKU catalog
- Manufacturing operations optimized using AI forecasting and workflow analysis
- Catalog optimized for large language model discovery
- Stronger positioning with major retail partners
- Full ownership of customer data across AI commerce channels
The Challenge: Infrastructure was limiting innovation
For years, POLYWOOD ran its ecommerce operations on a heavily customized Magento environment built to support its complex catalog and vertically integrated manufacturing model. The system worked at first, but it demanded constant engineering attention.
Over time, maintaining custom infrastructure became a full-time effort. Every new feature required significant development work, and even routine updates carried risk. Engineering resources that could have been focused on improving the customer experience or experimenting with new technologies were instead tied up maintaining the platform.
We spent hundreds of hours of development resources to build integrations with 3rd parties when other platforms already had it out of the box. A large percentage of the year was spent on platform upgrades. We were spending our energy on things that weren't creating value that needed to be done, but it wasn't really helping us win.
As the company grew, so did the operational burden. Supporting more than 150,000 product variations required extensive custom logic and ongoing maintenance. The platform that once enabled growth increasingly limited the organization’s ability to move quickly.
This constraint affected more than ecommerce operations. When engineering capacity is absorbed by maintenance, experimentation slows across the entire business. Exploring emerging technologies—including artificial intelligence—requires time, flexibility, and the ability to test new ideas rapidly. POLYWOOD had the ambition to innovate, but not the structural capacity to do so.
To move forward, the company needed a way to reduce operational overhead, reclaim engineering time, and create the conditions for sustained innovation.
Solution: Migrating to Shopify freed time for innovation and AI experimentation
When POLYWOOD migrated to Shopify, the team immediately experienced significant time savings.
Features that once required months of development were now available out of the box, allowing the team to focus on improving the customer experience instead of maintaining the platform.
We used to have people build features on front ends and back ends that already existed out of the box. Now I have the same people using it as a canvas. It used to be that I had a development team of 20 working on 'let's build a cart abandonment feature, let's build a wishlist feature.' Now I can use these resources to build a better customer journey—let's highlight the product features better, let's show why our texture of this finish is important.
With this newfound freedom, POLYWOOD now had something they never had before: the capacity to experiment.
With engineering resources no longer consumed by maintenance, teams across the organization began exploring how AI could improve the customer journey, automate repetitive work, and support decision-making. What started as small, informal experiments quickly expanded into broader adoption.
A new operating model enabled rapid experimentation
As experimentation spread across teams, POLYWOOD established a new structure to support rapid testing and adoption. Spiegel calls this “freedom within a framework.” The structure includes a weekly cross-functional AI standup where employees surface new tools, and IT evaluates them in real time. Spiegel explains how it works:
We purposefully don't have an agenda, which sounds counterproductive. But what happens is somebody says, 'I found this tool three days ago, I installed it on my home computer.' Then I have somebody from IT on the call who immediately figures out how do we run it, how do we make it safe. And two days later we have an approved way to run it here.
Adoption accelerated through practical automation
As more people shared tools and use cases, POLYWOOD learned that broad adoption grows from solving everyday problems, especially repetitive tasks that consume time but create little value.
We brought in somebody from our finance department and asked, 'What is taking you the most time?' The product renderings team—somebody opens a file from Solid Works, positions a camera, renders it out, moves the camera, renders it out, and does it again thousands of times. That is something so ready for automation. But often you have these old processes and you just never look at them.
Rather than focusing only on strategic projects, POLYWOOD encouraged teams to surface routine, manual work that could be automated.
Employees across departments identified processes that had long been accepted as part of daily operations but had never been revisited. In many cases, automation delivered immediate time savings and made the benefits of AI tangible across the organization.
Governance enabled safe, rapid AI adoption
POLYWOOD also built simple guardrails that allowed teams to experiment with AI freely while maintaining security and oversight.
The reality these days is that if you don't do that, people that want to will do it anyway. They're at home and they're going to use their own installation of ChatGPT and upload your company data. It happens all over no matter what you do. So you have to figure out how to adapt quickly but keep governance active.
To support experimentation while maintaining control, POLYWOOD implemented three practical safeguards:
- Teams were equipped with the tools they already preferred. POLYWOOD funded access to multiple leading AI tools, including Gemini, Claude, and ChatGPT.
- Dedicated hardware supported safe experimentation. The company repurposed older laptops connected to guest Wi-Fi and isolated from the corporate network. Employees could install local models and experiment freely without creating security exposure.
- Rapid approval became part of the operating rhythm. New tools surfaced during the weekly cross-functional AI standup, where IT evaluated them in real time and established approved usage quickly.
AI across the entire business
With the culture and guardrails in place, POLYWOOD has now deployed AI across virtually every function, including:
Development
The entire Shopify development team now uses AI-assisted coding tools like Claude Code. The team that used to spend months building a single 3rd party integration now spends virtually all its time on competitive differentiation.
Our entire development team today is using AI-assisted coding 100%. Now all of these people are working on new initiatives—the commercial portal, process improvements for customer service agents, self-service returns, self-service AI agents for customer questions.
Customer service
The team is experimenting with AI across customer touchpoints—phone, text, and chat—but Spiegel is deliberate about where AI adds value versus where it creates friction.
All the LLMs today are very conversational, and sometimes people just want to know where their order is. They don't want 'Hello Benjamin, how is your day going?' Sometimes they just want to know it will get there tomorrow.
Product discovery
The team is also piloting AI-based catalog discovery where customers describe their lifestyle (e.g., "I live by a lake, my family has four people, we love grilling, grandchildren come over often”). Once they’ve entered a description, they receive personalized product recommendations. For a brand with 150,000+ options, this moves the customer from browsing and filtering to a guided, conversational experience.
Manufacturing and operations
AI is being applied to demand forecasting, production scheduling, and warehouse optimization. The team uses video analysis to study traffic patterns inside their warehouse and determine the shortest paths for moving materials.
Shopify Sidekick
Across the organization, teams use Sidekick for assortment strategy, above-the-fold placement guidance, micro-app development, white-glove features, customer segment analysis, and metaobject management.
Results: Structural advantages from becoming an AI-first organization
By redirecting engineering capacity toward experimentation and embedding AI across the business, POLYWOOD positioned itself to move faster than competitors on the biggest shift in commerce since mobile: AI-driven discovery and purchasing.
Because AI was already embedded across the organization, POLYWOOD was prepared when AI-driven commerce began to emerge. That readiness has translated into several concrete advantages.
1. New AI leaders emerged across the organization
As engineering capacity shifted from maintenance to experimentation, unexpected AI leaders began to emerge across the company—often from outside traditional technical roles.
I would have not picked these people to be my top AI leaders. I would have thought it would be other people within my organization. Here's a fun thing: look at your own firewall data. Who pings ChatGPT 275 times a day? That person should probably be on your AI council.
At POLYWOOD, one of the key AI leaders doesn’t even have a technology job title. He simply enjoys experimenting and has automated his entire home using AI.
2. Catalog optimized for AI-driven discovery
Shopify was the first SaaS partner to bring POLYWOOD impactful AI capabilities—before they even asked for them. When AI-driven commerce began to emerge, Shopify provided tools to optimize POLYWOOD's 150,000-SKU catalog for large language model indexing and brought them into the ChatGPT shopping pilot early.
Out of all of our SaaS partnerships, you were the ones who leaned forward the most in enabling us to be ready for it. You came a year before that to us and said, 'Hey, we have a tool now that optimizes your knowledge base for LLM indexing, and here it is, and you only have to click this button, and it's done.'
3. Stronger positioning with retail partners
When POLYWOOD walks into meetings with major retail partners, the conversation is now completely different—in a good way.
It's a different posture when you're working with a Key retailer and either you have a conversation that is, 'Hey, we are improving our website and we're fixing the thing that broke last month,' versus, 'Hey, we are one of the five brands in this OpenAI pilot, here's some results from it, this shows how innovative of a brand we are.' Retailers want to work with brands that are doing that.
4. Owning the customer data
A common concern with AI commerce channels is the loss of brand control and customer data. With Shopify's integration, POLYWOOD retains full ownership.
When the orders come in, I get the same level of data I would get as if they buy it on the website. I have a flag in there that tells me it came from AI, but that's it. Everything else for us is the same. It makes it extremely easy to operationalize because it's not in a different order management system. Our customer service teams don't even have to worry about where the order was placed.
Advice to others building an AI-first organization
Spiegel’s experience offers a practical blueprint for leaders seeking to build an AI-first organization. Here’s what he recommends:
1. Create space before chasing innovation
When engineering teams spend most of their time maintaining systems and rebuilding baseline features, experimentation stalls. Leaders must first eliminate operational drag, so teams have the time—and mindset—to innovate.
2. Build guardrails, not roadblocks
Trying to ban AI rarely works. Employees will experiment anyway. The smarter approach is to create controlled environments where teams can test tools safely, with IT involved early and approvals happening quickly.
3. Surface unexpected AI leaders
AI champions don’t always occupy technical roles. Look for the curious people already experimenting with AI, automating their workflows, and pushing boundaries to lead AI implementation.
4. Start with unglamorous automation
Instead of launching moonshot initiatives, ask teams what repetitive work consumes the most time. Follow the pain. Automating manual processes creates an immediate impact and turns AI from theory into daily utility. People notice when you reduce or eliminate manual processes in their job, and it quickly grows organically from there.
5. Apply AI where it reduces friction
Not every interaction needs to be conversational or complex. AI should simplify experiences—not slow them down.
What’s next for POLYWOOD
With its infrastructure stabilized and AI embedded across operations, POLYWOOD is shifting from internal transformation to external differentiation. The company is now applying that foundation to reshape how customers manage outdoor living environments.
POLYWOOD’s AI roadmap for the next 6–12 months focuses on three areas:
- Interactive product visualization. Tools to help customers better imagine outdoor spaces, which is a unique challenge because, unlike indoor rooms with four walls, outdoor spaces are harder to visualize. This will include advanced renderings, interactive tools, and potential AR pilots.
- AI-powered catalog discovery. Expanding the natural language product recommendation pilot to help customers describe their lifestyle and receive personalized outdoor furniture suggestions at scale.
- Reducing post-purchase friction. Making warranty replacements, upgrades, and returns even easier through AI-powered self-service, while building a deeper CRM relationship over time.
