SAP Sapphire 2024 . Session BTP182
Border States: Gen AI Sales Assistant on SAP BTP
Border States Industries Delivered 2024
Field reps get answers in the conversation, not after it.
Presented at SAP Sapphire 2024
By the numbers
-
$4B+
Customer scale
-
130+ branches
Across 31 states
-
6th largest
US electrical distributor
Before
Before
- Rep on a customer call needs a part number, a substitution, or a price.
- Leaves the conversation to look it up across multiple systems.
- Thousands of SKUs across construction, industrial, utility.
- The answer arrives after the moment to act on it has passed.
After
After
- Generative-AI assistant grounded in Border States product and customer data.
- In-context answer, in the conversation, on the rep's actual screen.
- Built on SAP BTP with Joule and Business AI.
- Integration Suite pipelines from S/4HANA and Datasphere keep the answers current.
“Field reps get answers in the conversation, not after it.”
Built on
- SAP BTP
- SAP Business AI
- SAP Joule
- SAP Datasphere
- SAP Integration Suite
Why this matters
Border States co-presented this with Mindset at SAP Sapphire 2024 as the public reference for what a real Gen AI sales assistant looks like in distribution.
Border States Industries is the sixth-largest US electrical distributor — $4B+ in revenue, 130+ branches across 31 states, 100% employee-owned. They built a generative-AI sales assistant on SAP BTP with Mindset, and presented the work alongside us at SAP Sapphire 2024 in session BTP182.
The sales team's job is to know thousands of SKUs across construction, industrial, and utility markets. A field rep on a customer call shouldn't have to leave the conversation to look up a part number, a substitution, or a price. The assistant gives them an in-context answer in one place.
Built on SAP BTP using Joule and Business AI, grounded in Border States' own product and customer data through SAP Integration Suite pipelines from the underlying S/4HANA and Datasphere environments. Designed for the salesperson's actual workflow, not as a separate tool.
How we measured it
Session BTP182, co-presented by Border States and Mindset Consulting. The full architecture and roadmap is in the SAP Sapphire session library.
If we built this today
Concept · not delivered scopeSales answers, grounded in real order data.
This is a forward-looking concept, not the scope we delivered on this engagement. It is the build we would reach for now, grounded in SAP that ships today.
This engagement put a Gen AI sales assistant on SAP BTP so reps could get fast, trustworthy answers on products, pricing, and availability, and the modern build keeps that intent but grounds it in governed business data instead of a standalone bot.
The Joule agent
Sales Inquiry Co-pilot
Reads open sales orders, product availability, pricing conditions, and customer history, then drafts the answer or order action a rep needs in plain language. It proposes the next step and leaves the human to confirm.
SAP Joule, SAP S/4HANA Sales, SAP BTP, SAP Knowledge Graph · PROPOSE · Quote and order response time, plus first-contact resolution on sales inquiries
The Fiori app
Manage Sales Orders (Joule-embedded)
The standard S/4HANA sales order app with Joule embedded, so reps ask questions and trigger changes right where the order lives instead of switching to a separate assistant. Real-time availability and pricing come from the same screen.
Embedded in the Fiori launchpad
The data product
Customer Intelligence
Grounds the co-pilot in unified customer, order, and product data so its answers reflect actual buying patterns and current availability, not a stale snapshot. It gives the assistant a governed view of the account.
Intelligent Application on SAP Business Data Cloud
We would mine the real sales-inquiry process in SAP Signavio first, map the BTP and data landscape in SAP LeanIX, and let our MIND accelerators carry the old assistant logic over to the new Joule build.
A look at the work
Presented publicly
We took this work to the SAP stage.
The assistant listens to a live customer call, turns speech to text, then has a GPT model query SAP customer, sales, and invoice data and surface the answer in real time. Border States expects it to cut call-handling time by up to half by ending the hop across multiple SAP screens.
