Unigloves

Unigloves manufactures and distributes billions of disposable gloves annually, supplying hospitals, industrial clients, and healthcare providers across 50+ countries. When COVID-19 triggered an exponential surge in PPE demand, their 4-person order management team hit a hard scalability ceiling. Every purchase order — received as PDF attachments or inline email text — had to be manually keyed line by line into their ERP system, with complex orders containing up to 200 line items consuming several minutes each. The team fell progressively behind as volumes outpaced human capacity, delaying order confirmations, creating fulfillment backlogs, and eroding the high-touch customer service that differentiates Unigloves in a supply chain where order accuracy directly affects patient care outcomes.

Unigloves partnered with turian, an AI process automation vendor, to deploy a generative AI order processing agent embedded directly within their Microsoft Outlook environment via a native add-in. The agent uses large language models to read incoming purchase order emails and PDF attachments, extract all line-item data regardless of document format or layout, and write it directly into the ERP system — end-to-end, without human intervention for the majority of orders. Integration with existing ERP infrastructure meant no parallel database or new application was required. The Outlook-native deployment was deliberate: staff interacted with the AI through a tool they already used daily, compressing change management timelines and enabling rapid adoption across the small team. A human-in-the-loop review layer handles the minority of orders that fall outside the model's confidence threshold, preserving accuracy without creating a new bottleneck.

Processing time dropped from approximately 10 minutes to 2 minutes per order — a 79% reduction — with 85% of all incoming orders now handled end-to-end by the AI with zero human intervention. The team's effective order-handling capacity tripled without adding headcount, recovering roughly 84 hours of staff time per month — the equivalent of two full work weeks redirected toward customer-facing activity.

• 79% faster processing: 10 min → 2 min per order
• 85% full automation rate: most orders require no human touch
• 84 hours/month recovered for value-added work
• 99%+ accuracy, reducing downstream fulfillment errors
• Capacity sustained through staff absences that previously required escalation or overtime

• Target processes where manual data entry consumes a disproportionate share of a small team's capacity — order entry dominating 40%+ of a team's day signals strong automation ROI with a short payback period.
• Embed AI within existing tools rather than introducing standalone applications; deploying inside Outlook eliminated retraining overhead and accelerated time-to-value.
• AI as force multiplier: the same 4-person team tripled throughput without adding headcount, indefinitely deferring costly hiring.
• In healthcare supply chains, validate AI accuracy rigorously before full rollout — errors in order data can propagate to patient-facing fulfillment.
• Build for exception handling from day one: automating 85% of orders cleanly frees human judgment for the 15% that genuinely require it.