AI & Automation Case Studies

🎯 Challenge

At Convergys, one of the biggest frustrations was how long it took to track and resolve issues. Manual processes meant that issues sat unresolved for days, stakeholders weren't properly informed, and the team spent more time on administration than actual problem-solving.

The existing system relied on spreadsheets and email chains, making it difficult to:

• Track issue status in real-time
• Ensure proper stakeholder notification
• Maintain visibility across teams
• Generate meaningful progress reports

🔧 Solution

I implemented an automated workflow using Make.com (formerly Integromat) that connected our issue tracking system with automated notifications and reporting. The solution included:

• Automated issue assignment based on category and severity
• Real-time stakeholder notifications via email and Slack
• Escalation triggers for overdue items
• Daily summary reports for management
• Integration with project management tools for visibility

📊 Results

🎓 Key Learnings

This project taught me the importance of process automation in reducing human error and improving consistency. The biggest win wasn't just the time savings, but the improved stakeholder confidence that came from reliable, transparent communication.

🎯 Challenge

Weekly variance reports were draining more time than they were worth. The team was spending 4-5 hours each week manually analyzing data, identifying trends, and crafting reports that often missed key insights due to the volume of information.

The process involved:

• Manual data extraction from multiple systems
• Time-consuming trend analysis
• Inconsistent reporting formats
• Risk of missing critical variance patterns

🔧 Solution

I developed a prompt engineering system that leverages LLMs to automate knowledge capture and analysis. The solution includes:

• Automated data ingestion and preprocessing
• Custom prompts for variance pattern recognition
• AI-generated insights with confidence scoring
• Standardized report templates with dynamic content
• Exception highlighting for critical variations

📊 Results

🎓 Key Learnings

This project highlighted the importance of well-crafted prompts in achieving reliable AI outputs. The key was balancing automation with human oversight, ensuring AI suggestions were validated before inclusion in stakeholder communications.

🎯 Challenge

When exploring AI for project management applications, a critical question emerged: which LLM approach would deliver the best balance of performance, cost, and security for enterprise project management needs?

Key evaluation criteria included:

• Performance on project management tasks
• Data security and privacy considerations
• Cost efficiency for ongoing operations
• Integration capabilities with existing tools
• Customization potential for specific workflows

🔧 Methodology

I conducted a systematic evaluation comparing leading proprietary models (GPT-4, Claude) with open-source alternatives (Llama 2, Mistral) across real project management scenarios:

• Risk assessment analysis accuracy
• Stakeholder communication generation
• Project timeline optimization suggestions
• Requirements clarification capabilities
• Meeting summary and action item extraction

📊 Key Findings

🎓 Strategic Recommendations

The evaluation revealed that a hybrid approach offers the best value: using proprietary models for complex strategic analysis while leveraging open-source models for routine tasks like document processing and standard reporting. This approach balances performance, cost, and security considerations effectively.