Forty Use Cases, One Agent Layer
Kale Holding is one of Turkey's largest industrial groups. Kale Seramik makes ceramics at scale; Kalekim makes building chemicals. Both run on SAP, both carry decades of product, contract and operational history.
The question was never whether AI could help. It was where to start. Instead of one pilot, Kale mapped the whole landscape: over 40 candidate use cases, grouped into six domains, spanning operations, finance, commercial and IT.
The Use Case Landscape
Over 40 use cases grouped into six domains across Kale Seramik and Kalekim. Each one reads from the same governed layer, so the answers agree with each other.
Supply chain & inventory
Dead stock moved before it ages, orders proposed before they are asked for.
- Dead stock transfer between plants and warehouses
- Potential order generation from sales patterns
- Dispatch and vehicle load optimisation
- Demand forecasting by product family
- Stock segmentation and coverage analysis
Finance & controlling
Controllers ask in plain language instead of waiting on the next report cycle.
- Budget versus actual monitoring
- Profitability reporting by product and channel
- Financial statement analysis
- Reconciliation questions answered from source data
Pricing & packaging
Commercial teams price from live stock and sales data, not last quarter's sheet.
- Optimum sales price suggestions
- Product positioning against the portfolio
- Customer-specific package drafts from stock and sales history
Product information access
One agent answers from PIM, SAP and the document library at once.
- PIM product attributes and specifications
- SAP stock, price and lead time
- Technical documents and datasheets
- Reachable from web, the B2B portal and Teams
Contract & document review
Every draft checked against the template before it reaches a signature.
- Version comparison across drafts
- Deviation from approved templates
- Date and term checks
- Risk detection on purchasing, sales and service contracts
IT operations & audit
The first line of support answers itself, and audit evidence collects itself.
- L1 ticket classification and routing
- Knowledge base answers for common requests
- Configuration and security evidence collection
- ISO 27001 audit preparation
What Makes It Work
One Foundation Under Every Use Case
Forty use cases do not mean forty projects. SAP transactional data, PIM product attributes and the technical document library land in the same governed lakehouse. Each new use case is a question asked of data that is already there.
Stock, price and lead time carry one definition across both companies. A pricing agent and a finance report reading the same number is not a coincidence — it is the same column.
Agents Where People Already Work
The same product agent answers on the public site, inside the B2B portal and in Teams. Dealers, sales engineers and internal staff each get the slice they are allowed to see — one agent, one permission model, three front doors.
Documents Treated as Data
Contracts, datasheets and audit evidence are indexed alongside structured records. A deviation from a template, a missing renewal date or a configuration proof for ISO 27001 is a query, not a week of reading.