Back to Experience

Featured Project

GenAI Cost Forecasting Model

A predictive system that transformed AI governance at Sherwin-Williams by forecasting implementation costs before deployment—reducing approval cycles by 50%.

50%

Faster Approvals

64,000+

Employees Impacted

6 weeks

Development Time

Enterprise

Adoption

The Story

During my internship at Sherwin-Williams, I noticed a critical bottleneck in the AI governance process. Business units would submit requests for AI tools, but estimating costs was a manual, time-consuming process that delayed approvals by weeks.

Inspired by ML and economic forecasters who predict product costs for future planning, I asked: "Why couldn't we forecast the cost of implementing AI tools the same way?"

I authored a white paper outlining the concept and pitched it to my manager, David Wendt. He saw the potential and gave me the green light to develop it further while balancing my other AI governance responsibilities.

Working tirelessly with ML engineers to validate the approach, I developed a presentation and presented to both my direct manager and the VP overseeing all AI initiatives at Sherwin-Williams. They loved it.

Within six weeks, I led the team that brought this initiative from concept to production. The tool fundamentally changed how Sherwin-Williams evaluates AI investments and was adopted enterprise-wide.

How It Works

Unstructured Business Inputs

Natural language requests from business units describing their AI tool needs

We need a chatbot for customer service
Want to use Copilot for our sales team of 50
Looking to implement document summarization
Need AI for inventory forecasting

Data Transformation

Input: Business Request

"Hi, our marketing team of about 50 people wants to start using Microsoft Copilot for creating presentations and analyzing campaign data. We'd need it integrated with our existing SharePoint and Teams setup. What would this cost us annually?"

Copilot Studio LLM

Extracts, classifies, and normalizes parameters

Output: Structured Data

tool: "M365_Copilot"

users: 50

department: "Marketing"

integrations: ["SharePoint", "Teams"]

use_cases: ["presentations", "analytics"]

tier: "Enterprise"

Technologies Used

Microsoft Copilot StudioAzure AI ServicesPower BIPythonLLM/GPT ModelsAzure Cost ManagementPower AutomateSharePoint

Key Outcomes

Reduced AI proposal approval cycle time by 50%
Adopted enterprise-wide across Sherwin-Williams
Enabled data-driven AI investment decisions
Presented to and approved by VP of AI
Fundamentally changed AI governance process
Supported ethical and cost-effective AI adoption for 64,000+ employees