Work / Case study 2 of 3
Process & automation · SAP ECC · TC Energy, Calgary, 2016–2020
Redesigning and automating procure-to-pay in SAP
Procure-to-pay is where purchasing, suppliers and finance meet inside the ERP, and one of the first places enterprises are putting AI agents to work. For four years I worked and improved it at national scale in SAP ECC for a leading North American energy infrastructure company, first as a Senior Buyer, then as the Canadian procure-to-pay analyst. I cleaned the data, measured the process from the client's side and automated the steps that did not need a person.
Context
TC Energy, a leading North American energy infrastructure company, runs high-volume procurement for operations and capital projects on SAP ECC. I joined Supply Chain Purchasing Canada in Calgary as a Senior Buyer (May 2016 – Dec 2019), then became the Supply Chain Procure-to-Pay (Canada) Business Analyst (Jan – Nov 2020), owning analytics and SAP data for both the Canadian purchasing team and Accounts Payable.
The business problem
High-volume procure-to-pay accumulates friction: inactive materials and vendors in the master data, purchase orders left open long after the work is done, invoices that fall into exceptions, approvals that depend on manual hand-offs, and KPIs that describe the process from the inside rather than from the client's side. Each one costs buyers, Accounts Payable and the business time, and each one makes the data less trustworthy for decisions.
My role
As a buyer, I worked the process every day and took on its data problems. As the procure-to-pay analyst, I owned analytics and SAP data for the Canadian purchasing and Accounts Payable teams, and led the business requirements for a PO-closure automation project as its main point of contact.
Key decisions
- Clean the foundation first. I cleaned the material and vendor masters, removing inactive records so the process ran on current data.
- Focus effort with evidence. I built a vendor PO volume and frequency report with ABC analysis to target vendor contact reviews where they mattered most, and analyzed the eCommerce implementation with ideas for improvement.
- Measure what the client experiences. I designed a new, client-focused view of the PO cycle-time KPI, built an Accounts Payable on-time-pay history report covering 2014–2020, and ran the first performance-measurement survey for the Canadian purchasing team.
- Clear the backlog safely, then automate it. I ran Canada's first SAP mass purchase-order closure with zero business impact, then defined the business requirements to automate PO closure so it would never depend on manual effort again.
- Automate the routine hand-offs. I built the purchasing approval flow in Microsoft Power Automate.
Outcome
- Leadership gained the data and survey feedback it needed to sharpen its KPIs and decisions.
- Cleaner master data and fewer invoice exceptions, recognized by peers.
- The backlog of open purchase orders cleared without disrupting the business, with automation defined to keep it clear.
- Nine peer recognitions in 2018–2019 for speed, detail and SAP knowledge.
Why it matters for AI
This is the work that comes before agents, and the step that decides whether a procure-to-pay agent ever leaves the pilot. An agent can only act safely when the master data is clean, the process is clearly defined, the KPIs measure what the client actually experiences, and approvals and controls are explicit. Rule-based automation, such as approval flows and PO closure, is the first rung; agents are the next. I have already built the first rung inside SAP ECC, so I design agents that fit the way the process actually runs, not the way a slide says it does.