Skip to main content

Bringing scale and precision to directed spend benefits with AI

Take a look at the model driving efficiency, speed and item-level control at the point of sale for millions of members.

Sam Garbrecht, Senior Director of Directed Spend Product, Optum Financial | July XX, 2026 | 4-minute read

From static lists to dynamic intelligence: Evolving directed spend

When a member uses their benefits at checkout, there’s no room for uncertainty. The right items need to be covered instantly and the experience needs to feel seamless.  

Behind that moment is a complex system — managing more than 14.4 million product codes, constantly evolving retail inventories and strict benefit rules that must be applied accurately every time.   

Optum Financial brings this together through AI-powered Approved Product Lists (APLs) that continuously evolve across our expansive S3® network of over 70,000 retail locations nationwide that helps plans deliver a more precise, reliable and scalable directed spend experience.  

What powers a smarter checkout experience

Instead of relying on static lists, our approach at Optum is designed to keep pace with a dynamic product ecosystem. It combines automation, human expertise and continuous learning to ensure eligibility decisions are both fast and accurate. 

Three ways this comes together

  1. Confidence at checkout — Each item in a member’s basket is evaluated at the product level in real time, helping ensure eligible items are covered and ineligible items are excluded without disrupting the experience based on each member’s unique benefits. 
  2. An AI model that continuously learns — The Optum AI model categorizes the majority of products in the software automatically, while flagging exceptions that can’t be determined for expert review. That feedback is then used to refine and improve the model. 
  3. Operational scale without added friction — Automation reduces the need for manual processing, making it easier to onboard new products — such as seasonal items — and maintain accuracy as product catalogs grow.

How a product becomes eligible

Behind every approved purchase is a continuous cycle that keeps the system accurate and up to date. 

When new product data enters the software, our APL AI model evaluates available attributes — such as item descriptions and retailer-provided information — to determine the most appropriate category within the benefit taxonomy such as over the counter, healthy food, or utilities.

Products with high-confidence matches can move forward immediately, while those with lower confidence are routed to internal experts for validation. These decisions are not isolated — they feed back into the model, improving performance over time through ongoing retraining.  

This balance of automation and human oversight allows the system to scale efficiently while maintaining consistency and control.

Designed for real-world complexity

Directed spend operates in real time across thousands of locations where product assortments are always in flux.  

That’s why this approach is designed to:  

  • Support mixed baskets where only specific items are eligible in a single transaction.  
  • Keep pace with new product introductions from a growing retail network.  
  • Reduce checkout disruptions by ensuring products are recognized and correctly categorized.  
  • Maintain alignment with benefit rules through consistent classification standards.  

Why it matters for plans and members

At its core, this capability helps ensure that benefits work the way they were intended. 

For plans, that means greater confidence in how dollars are spent — with more precise controls applied at the item level. 

For members, it means a smoother experience at checkout with their supplemental benefits card, with fewer surprises and more reliable access to covered products. 

This combination of precision and experience helps build trust in the benefit itself.

Continuing to evolve with the market

As product ecosystems expand and benefit designs evolve, maintaining accurate eligibility at scale will only become more important. 

Today, our APL model can automatically tag over 90% of incoming items, significantly reducing manual review and helping products become eligible faster. 

We are continuing to enhance our models, expand product coverage and refine how data is integrated and categorized — helping ensure that directed spend programs can adapt alongside the needs of plans and members.

Learn more about our offerings

Join Our Growing S3 Merchant Network

The S3® merchant network (formerly Solutran) brings millions of motivated shoppers to your stores through pre‑funded health benefit dollars with seamless acceptance at checkout.  

Supplemental Benefits for Members

Healthy Benefits+ is a smarter, simpler way to manage supplemental benefits. We help health plans maximize every benefit dollar while delivering a modern, member‑first experience.

Related healthcare insights

Case study

Modernizing provider review for consistency and compliance

Explore real-world examples of how hospitals are streamlining provider review workflows and supporting quality oversight with Crimson AI.

Article

Policy watch 2026: New laws and regulations impacting pharmacy

With new federal rules proposed and 1,500 PBM-related bills active in state legislatures, the scope of pharmacy regulation is changing.

Video

Optum HouseCalls visits can improve outcomes: Delores’s story

Optum HouseCalls offers in-home or virtual clinician visits at no extra cost. Watch Nurse Practitioner Dyan guide Delores’s care journey.