Scaling BPO success with agentic AI: Real-world client wins -
Register now
How the beauty retailer builds AI agents to scale enterprise operations for 3.5M+ subscribers
When Sree Sreedhararaj is in a room full of CTOs, the conversation sounds nothing like it did three years ago.
"Two, three years ago, when I was in a forum of CTOs and CIOs, the conversation was around digitization, digital transformation, or cloud journey," explained the Chief Technology Officer at IPSY, a beauty subscription service with 3.5 million customers. "Now the conversation is, ‘What are you doing with AI, and how are you doing it?’”
For retail leaders overseeing complex operations at massive scale, AI has simply become essential. In our latest webinar with Pinkfish CEO "CK", Sree shared how IPSY piloted this transition, and what other retail execs can learn from their experience.
Read a summary of our chat with Sree below, and watch the recording at any time.
Sree used an analogy to describe how AI accessibility has changed. Traditional AI used to be "like the cockpit of a plane, where you have thousands of buttons, signals, and analytics," he explained. "Only a pilot and people who have gone through training recognize what that is."
Today’s AI platforms operate more like a Tesla. "You have unlimited buttons. You have a screen. Most things are controlled by autopilot, and a lot are just one-click. Now, chances are a lot more people are willing to sit in the pilot’s seat and take off."
The democratization of AI creates both opportunity and risk. Making everyone a pilot without proper training can be dangerous, which makes AI governance, ethics, and empathy more important than ever.
IPSY’s AI journey began with digital operations, specifically the manual updating of product information and pricing across hundreds of items on weekly and monthly cycles.
"Our product velocity is super fast. We churn through different assortments on a weekly, sometimes monthly, basis," Sree noted. "It requires us to refresh product pricing and features on a regular basis. It used to be humanly impossible to update 800 to 1,000 products at once."
Writing custom code for each update took up valuable engineering time. Instead, IPSY used Pinkfish to automate the handoff between merchandising teams and digital operations. The results exceeded expectations—not just automating the process, but making it adaptive and communicative.
"Pinkfish easily integrated with Slack, communicating back to business teams the updated price," Sree explained. "That created this increased adoption of AI across the company."
IPSY’s success with pricing opened doors to more ambitious AI use cases. The retailer expanded into product page descriptions, validations, and brand creative work. But rather than replacing individual roles, they redeployed talent to higher-value activities.
"Did we replace people with it? We haven't, because we moved those people into other areas where we wanted capacity to do additional, human-related things," Sree clarified.
And it’s consistent with Sree’s broader philosophy about AI adoption. Instead of asking "where can we use AI," he recommends focusing on "the things that slow us down, from a systems, process, or people perspective—wherever we spend a lot of time in mundane tasks."
Sree’s team made a strategic decision to consolidate automation tools rather than deploying point solutions for each department. With existing integrations to systems like Kineo for product information management (PIM) and their commerce platform, expanding Pinkfish’s role made more sense than adding new vendors.
"With the advancements of Pinkfish and its integration with our systems, why wouldn’t we just automate the whole process?" Sree asked.
IPSY’s technology consolidation reflects a broader shift among retailers. Rather than managing dozens of specialized tools, platforms like Pinkfish enable anyone to build solutions faster and more intuitively than RPA and iPaaS—popular, but outdated legacy choices.
Sree drew important distinctions between RPA and AI-powered automation. RPA solutions become brittle when systems change and require specialized engineers to update scripts and workflows.
"With agentic AI, automations are adaptive in that you inform the goal or objective," he explains. "If you guide the AI agent, then it will learn as the world changes, adapt accordingly, and execute."
When exceptions occur, agentic systems alert users and incorporate new guidance, leaning on self-healing workflows rather than fragile scripts.
Asked about his “magic wand wish” for AI in retail, Sree immediately pointed to supply chain. "Supply chain is one of the most complex, least understood, and least-innovated spaces still," he argued.
Consumer expectations have moved far outside of four-wall stores. “Retailers are optimizing for completely different priorities now, but supply chain innovation hasn’t caught up.”
Areas like inventory planning, forecasting, delivery route optimization, and dynamic cost management represent massive opportunities for AI automation, considering their impacts on customer experience and brand perception.
Sree offered tactical guidance based on IPSY’s experience: Start with operational efficiency, not disruption. "Don’t think, ‘How do I invest millions of dollars and change the world?’ It’s not going to happen overnight."
Focus on governance and values first. "Create what matters, what values you stand behind, and make sure you’re not breaking that rule when applying AI."
Perhaps most importantly, Sree emphasized changing how you think about AI performance metrics. "Don’t look at agentic AI from a people replacement perspective. Look at it from a skills empowerment perspective."
For IPSY, the journey continues to expand across geographies, languages, and use cases. But the foundation remains consistent: start small, prove value, maintain human focus, and scale thoughtfully.
As Sree put it, AI is no longer a choice for retailers. "Either absorb it and react to it, or step away and say that it’s just hype. But in reality, it’s here to stay."
Ready to automate your retail ops? Start building AI agents and open new doors for AI use cases across your organization.
Expanding possibilities for people at work
