AI customer intelligence agent for a global haircare brand

Client

A global leader in the haircare industry, renowned for its patented bond-building technology designed to repair and strengthen damaged hair

Industry

Retail, FMCG, beauty

Services

AI consulting, Gen AI consulting, custom software development, data engineering, data platform consulting

Tech stack

Snowflake, Snowflake Cortex AI, Streamlit, Python

Challenge

A global haircare company, which is known for its scientifically proven products, faced a common yet critical business challenge: its customer feedback was scattered across numerous online platforms, including Sephora and Trustpilot. Despite the immense potential of this data, its unstructured nature necessitated manual analysis.

Barriers to Efficient Customer Insight

The company's marketing and product teams lacked a common understanding of customer sentiment, which prevented them from responding to critical business questions without significant manual effort. This resulted in several major pain points:

●

With reviews spread across multiple sources, there was no efficient way to consolidate feedback or identify overarching trends

●

The inability to automate analysis meant that shifts in customer satisfaction—whether positive or negative—could go unnoticed until they impacted the brand’s revenue or reputation

●

Without review data, segmenting customers into distinct personas (e.g., value-conscious buyers vs. salon professionals) depended solely on internal information, preventing the client's team from tracking product satisfaction levels across customer cohorts

●

The client had access to powerful data platforms like Snowflake but needed a strategic technology partner to leverage its built-in AI capabilities to drive business value

Recognizing the need for a modern data strategy, the company continued its long-term partnership with ITRex. Having previously collaborated on a successful data management and analytics initiative, the client trusted our team to build a solution that could unlock the full potential of their customer feedback.

Solution

ITRex created a proof of concept—a powerful AI-powered review analysis agent—to demonstrate the capabilities of modern AI technologies without the overhead of a large-scale project. We designed the solution to ingest, analyze, and visualize customer feedback in near real time. For this, we utilized Snowflake Cortex AI and created a user-friendly interface with Streamlit, an open-source Python framework. The platform empowered non-technical business users to:

●

Automatically pull customer reviews and metadata from disparate sources like Sephora and Trustpilot into a single, secure Snowflake environment

●

Use Snowflake Cortex AI to perform sophisticated analysis, including sentiment analysis, trend forecasting, and automated summarization of products’ pros and cons

●

Identify and understand distinct customer personas based on their feedback and purchasing behavior

●

Explore data, view KPIs, and gain actionable insights using a simple web interface, all without involving the data team

Our team followed a streamlined, four-step process to quickly move from concept to a tangible demonstration of value:

Discovery & goal alignment. We kicked off the project by aligning with stakeholders on the core business challenges and defining clear goals for the PoC. The primary objective was to demonstrate the AI could be enabled from day one, utilize Snowflake AI Capabilities (Cortex AI), and validate that these capabilities could deliver actionable insights securely and efficiently.

Rapid prototyping & development. We built an end-to-end AI agent and analyzer that included a custom parser for review extraction, a Snowflake data lake for secure storage, and Cortex AI integration for natural language processing tasks.

Intuitive UI for business users. Using Streamlit, we built a simple and secure web interface, allowing stakeholders of all levels and technical prowess to interact directly with the AI-generated insights.

Demonstrating business impact. Through a live demo, we showcased the platform’s ability to generate KPI dashboards, gauge sentiment, segment customers, and produce intelligent recommendations, proving the immediate business value of the solution.

AI-Customer-Intelligence-Agent-for-a-Beauty-Company

Tech

The solution was designed as an end-to-end AI pipeline that functions entirely in the client's Snowflake and AWS environments. By using Snowflake's native AI capabilities, the ITRex team managed to improve data governance while reducing development overhead.

●

Secure AI pipeline The AI agent's "brain" is a Snowflake-powered data storage and Python-based processing pipeline that ingests raw data, processes it, and analyzes customer feedback entirely within the client's secure Snowflake environment. These pipelines can glean insights from both structured (e.g., star ratings) and unstructured (e.g., text comments) data from various sources. Snowflake ensures isolated data processing, which means that all AI computations take place in a dedicated environment, without interference from external or globally connected models. No data is transmitted outside the network at any time, ensuring complete privacy, security, and regulatory compliance.

●

Cortex AI for zero-overhead NLP Instead of building and managing a complex MLOps infrastructure, our team utilized Snowflake Cortex AI's managed, serverless feature to deliver complex insights. We called AI models provided by Snowflake to perform key tasks directly on the raw text data: ● Classifying reviews and tracking customer satisfaction trends over time ● Automatically creating concise "Advantages" and "Concerns" for each product, distilling hundreds of reviews into actionable bullet points ● Classifying users into distinct personas based on the content of their feedback By using the "low-code" approach to AI development, we were able to quickly prototype the customer intelligence solution and demonstrate the power of enterprise-grade LLMs without incurring the cost and complexity of implementing custom models.

●

Architecture overview ● Data platform & AI engine: Snowflake ● AI & ML framework: Snowflake Cortex AI ● Data ingestion: Custom Python scripts ● User interface: Streamlit framework ● Deployment environment: Secure, client-hosted environment ● PoC deliverables: Fully functional Streamlit web application, AI-powered KPI dashboard, automated product summary reports, and a trend forecasting module