Azure AI Procurement Transformation for Global IT Firm

About the Client 

The client is a global IT services company that functions in North America, Europe, and Asia-Pacific. The business runs large-scale purchasing operations, processing thousands of Purchase Requisitions (PRs) each year for infrastructure, facilities, IT gear, and other operational necessities.

The client had mature ERP and procurement systems, but the process of creating and consolidating PRs was still very manual. This led to fragmented sourcing, missed chances for consolidation, and poor vendor negotiations

Project Objective 

The goal was to use Azure AI Foundry Services to create an AI-powered procurement intelligence platform that could:  

• Analyze and understand PR descriptions that aren’t structured. 
• Smartly put comparable PRs together, such as power cables, networking equipment, and consumables.  
• Combine several PRs into one optimized RFP. 
• Make sourcing more efficient and enable more power in negotiations.  
• Work well with the procurement processes that are already in place  

The client required an AI system that was safe, easy to understand, and ready for business use, not just a chatbot or proof-of-concept. 

Project Details

Bloom was hired to create and put into action a solution that used Azure AI Foundry Services to modernize procurement procedures. This solution included large language models, embeddings, and orchestration workflows.

The solution focused on smart text interpretation and matching, which helped procurement teams find similarities between PRs, even when multiple teams or areas wrote the descriptions in different ways.

Some of the main areas of focus were:

• AI-based PR classification and clustering. 
• Finding semantic matches in text and comparable materials. 
• Creating drafts of RFPs automatically. 
• Validation for governance with a human in the loop. 

Engagement Model

• Procurement Process Discovery: Learn about the PR lifecycle, sourcing workflows, and problems. 
• AI Architecture Design: A reference design for Azure AI Foundry Services that includes security and governance. 
• Model and Prompt Engineering: Smart text matching and logic for combining text. 
• Integration and Workflow Automation: Linking ERP and procurement systems. 
• Operational Enablement: Monitoring, controls, and ongoing improvement. 

Overview of the Azure AI Foundry Solution

Core Capabilities Implemented

1. Smart Text Matching and PR Consolidation

1. PR descriptions taken from purchasing systems 
2. The Azure AI Foundry Services embeddings were utilized to turn PR text into semantic vectors. 
3. Using AI to score similarities to find related materials. 
4. Automatic grouping of PRs (e.g., multiple power cable requests across locations). 
5. Suggested combined RFPs for teams that buy things 

2. Building AI-powered RFPs

• We employed Large Language Models (via Azure AI Foundry Services) to: 
  • Normalize material descriptions, 
  • Extract technical specifications, 
  • Generate draft RFP content. 
• Before being released, procurement teams look over, improve, and approve them.

3. Governance and Explainability 

• Confidence scores and descriptions of how similar each group is. 
• Checkpoints for human approval before consolidation. 
• A full record of AI decisions and modifications. 

Technology Highlights

• Azure AI Foundry Services for managing the lifecycle and orchestration of models. 
• Azure OpenAI models for understanding and creating language. 
• Semantic search based on embeddings for smart text matching. 
• Azure SQL and Azure Blob Storage for storing PR and RFP data. 
• Azure Functions and Logic Apps for managing workflows. 
• Microsoft Entra ID for providing safe access and controls based on roles. 

Key Challenges

Technical Challenges 

• PR descriptions that are very different from one region to another. 
• Similar materials that are described in various ways. 
• Avoiding false-positive consolidation. 
• Ensuring low-latency AI responses at enterprise scale. 

Non-technical Challenges

• Procurement team trust in AI-driven recommendations 
• Managing change in existing sourcing processes. 
• Making sure that audit and compliance teams can see all. 

We addressed these issues diligently by using semantic modelling, explainability, and human-in-the-loop controls. 

Business Value delivered by Bloom

• Used Azure AI Foundry Services to develop a useful, production-ready AI procurement solution. 
• Put AI into current procurement processes. 
• Governance and controls that are balanced with automation. 
• Allowed procurement teams to focus on strategy instead of doing manual analysis. 
• Created an AI foundation that may be used again for future sourcing use cases. 

Market Significance

The procurement system powered by Azure AI Foundry Services made it possible for the client to: 

• Make sourcing more efficient across global operations. 
• Improve vendor negotiation leverage through consolidated RFPs. 
• Reduce procurement cycle times. 
• Make auditing and transparency stronger. 
• Make procurement a strategic and data-driven role. 

Delivery Minutes

• Mode of delivery: Hybrid (remote delivery including workshops for stakeholders across the world). 
• Time frame: 12 weeks (from discovery to production rollout).  
• Team: AI Architect, Azure Engineer, Prompt Engineer, Procurement SME, and Engagement Manager. 
• Governance: Weekly steering meetings and permissions depending on milestones. 

Results 

Within 4 to 6 months of implementing Azure AI Foundry Services, the client achieved:  

• 30–40% reduction in procurement cycle time. 
• 25% increase in PR consolidation opportunities identified 
•  Better cost savings by combining bigger and combined RFPs. 
• A big drop in the amount of work needed to manually review PRs. 
• The procurement team became more confident because AI recommendations made sense.

With Azure AI Foundry Services, the client turned scattered procurement data into useful information that enabled faster decision-making, better negotiations, and a more unified global sourcing strategy.