AI Embeds Itself in Operations

Inside the AI Transformation of Parts, People & Processes

Jun 25, 2026

Hi, I’m Lily. I live in the world of distribution, where the real work happens at the task level: placing a store order, querying an ERP screen, finding the right product mid-conversation. AI used to sit above those moments. This week, it’s inside them.

When AI lives where the decision happens, the gap between knowing and doing collapses. That’s the shift worth watching: AI as operational infrastructure.

If you only read one thing this week, it is this:

Grocery Outlet and AI ordering platform Afresh have deployed store-level intelligence across all departments — fresh, center store, general merchandise — at roughly 550 stores. Store workers now spend half the time on ordering. Average sales across Afresh partnerships rise 3% and shrink drops 25%. On a discounter’s margins, those three numbers are the whole business case.

What’s Working in the Field

Epicor AI Agents Enter Distributor ERP

Epicor launched Prism across the UK and European markets, embedding 18+ vertical AI agents directly into Prophet 21, the ERP used by thousands of distributors, and Kinetic, its manufacturing platform. The agents surface supply, demand, and fulfillment risks from live ERP data without requiring separate analytics tools or additional modules. For distributors, that means planner productivity without a detour: Prism pulls context from the system of record and delivers it in conversation, right inside the tool they already use all day. Epicor reports Prism Developer cuts ERP customization time by 60%. AI isn’t being layered onto the ERP anymore. It’s moving inside it. Read more here

Grocery Outlet’s AI Halves Order Time

Grocery Outlet has deployed Afresh’s AI ordering platform across fresh, center store, and general merchandise departments, making it the first retailer to use Afresh’s full-store, multi-category system. The AI generates store-specific recommendations from local demand patterns, flags inventory for human review, and extends real-time visibility from corporate teams down to the individual item. Store workers now spend half the time on ordering. Across Afresh partnerships, average sales rise 3% and shrink falls 25%, with a 94% adherence rate against planned orders. On a discounter’s margins, those aren't incremental gains, but rather structural ones. Read more here

3M AI Arrives at the Counter

3M officially launched Ask 3M, a conversational AI assistant that gives industrial users direct access to 3M technical expertise, starting with adhesives and tapes across 49 technology platforms. Built on AWS Amazon Bedrock and AgentCore, the tool surfaces product recommendations grounded in verified 3M documentation, then links directly to authorized 3M distributors for purchase. Users can download source documents in-session without leaving the conversation. The scale of the knowledge base is the foundation. Manufacturers are using AI to close the gap to the end customer, and routing that customer to authorized distributors rather than around them. That’s a new kind of inbound: one initiated by the manufacturer’s AI instead of a sales call. Read more here

What’s In My Ears

Riding the Next Upcycle: Jeff Newell on AI, Data Centers and Demand

In this episode of Supply Chain Connect, Jeff Newell, President of Mouser Electronics, maps how AI-driven data center demand is powering the next upcycle in electronic component distribution. Topics worth your time: how “golden screw” memory constraints are choking certain supply chains right now, how Mouser is investing in AI tools and warehouse automation to handle shifting lead times, and what Newell thinks will separate winning distributors from the rest when the cycle peaks. Concrete, forward-looking, and grounded in Mouser’s actual operations. Listen here

Lily’s Quick Take

What connects AI agents inside Epicor’s ERP, Afresh’s ordering platform at Grocery Outlet, and 3M’s AI assistant routing customers to distributors? None of them are standalone tools that operators must go find and open separately. They’re embedded at the point where decisions already happen: in the ERP screen, at the order terminal, during the product lookup. AI is closing the distance between a decision and the system that executes it.

The implication is practical: the next AI investment worth making isn’t a separate platform sitting alongside your workflow. It’s the intelligence woven into the systems you already use every day.

Operational complexity doesn’t shrink on its own, but intelligence embedded in the seams of your systems finally makes it manageable.

Until next week—keep your systems learning!
— Lily @ InstaLILY AI