In healthcare, accounts payable (AP) and supply chain are in constant motion together. Every invoice connects to a purchase order, a vendor, and a delivery—and keeping that chain moving requires both teams working from the same information.
And that volume is real, with paper invoices, EDI transactions, and purchase orders all arriving in different formats and on different timelines.
Yoga for FSM was designed to handle all of it.
Chad Tucker, RPI’s Yoga Software Vice President will walk through how Yoga for FSM brings the full invoice lifecycle into a single, connected process built for healthcare AP and supply chain teams. Session attendees will see how it:
- Manages the full invoice lifecycle, from ingestion through posting
- Uses built-in dashboards and KPIs designed specifically for healthcare AP
- Handles EDI transactions without manual intervention
- Connects AP and supply chain teams around a visible, trackable resolution process
- Leverages intuitive interfaces, customization options, and Infor FSM integrations
If your team sits in AP, supply chain, or financial systems management and you’re tired of invoices getting stuck between teams, you won’t want to miss this session.
Transcript
Chad Tucker
Good morning. Thank you for joining this webinar today to talk about RPI Consultants’ Yoga flexible software — our intelligent AP automation solution.
My name is Chad Tucker. I’ve been with RPI for a little over 11 years — November will be 12. But I’ve been in this space for over 15 years, working at different hospital systems before joining RPI. For essentially the past 15 years, I’ve been doing nothing but nerding out on AP automation solutions and obsessing over the gaps between capture and what AP teams actually need. Before developing our own software, we implemented solutions for a variety of different partners, so we have a clear picture of what works, what doesn’t, and what AP teams are really looking for — and we’ve made sure to bake all of that into this solution.
I also work with the Institute of Finance & Management to make sure we’re all AP-certified solution consultants, and we’re all Infor certified as well. My own start in the procure-to-pay space was in the Marine Corps, working for the Office of the Comptroller of the Second Marine Logistics Group — doing large-scale, high-volume, globally distributed P2P. That started the journey to where we are today.
Yoga has a couple of different solutions. Today we’re mostly going to be talking about our AP invoice automation solution. We also have a solution for EDIs, a solution for scanning packing slips, and a statement matching automation solution that we’ll be releasing at the start of Q3 — it’s in beta right now, and I’ll show a quick screenshot of what that looks like later.
I also want to highlight that we’re a Databricks partner. Our backend data lives in Databricks, which makes it easy for us to share that data with your enterprise data warehouse or allow you to run your own reporting tools on our raw data. So you have access to all the data we’ll be showing here today.
For AP automation, we’re very Infor-focused — RPI is an Infor partner. We support both Infor CloudSuite and version 10, and there are some differences between the two. CloudSuite has more built-in functionality for document storage and approval routing, but Yoga still handles invoice capture, intelligent data extraction, and all the typical header validation: does the vendor exist, does the PO exist, do the PO lines exist, and so on. Exceptions and workflows are split — the ones that primarily prevent you from creating an invoice get stopped in Yoga, while things like buyer messages and match exceptions stay in Infor, because at that point the invoice is already in the system and it’s a matching issue.
For Lawson version 10, Yoga can handle the entire end-to-end invoice stack. Infor still comes into play for match exceptions via MA126. For approval routing in version 10, you have two options: Yoga has a very flexible approval system, but Lawson version 10 also has a Landmark application called APIA — AP Invoice Approval. If you build your approvals in APIA, they transition to CloudSuite when you make that move. When it comes to change management and how often you want to present change to the managers who approve invoices, we can thoughtfully work through whether you should do approvals in Yoga or consider APIA for future-proofing. We walk through all of that as part of our consulting process.
Everyone hears a lot about AI today, and invoice processing has been using AI and machine learning for a long time. There’s definitely been some novel technology that applies to invoices, and we’ve carefully vetted what to do and what not to do. It’s very easy to find companies that have taken a generic commodity OCR tool, plugged it into ChatGPT, and called the result invoice automation. But those systems are deterministic — they’re not going to be 100% accurate, and with invoices you want 100% accuracy. Within the first 90 days of go-live, we can typically achieve touchless PO processing of upwards of 60%.
We use AI to help with the long tail — the edge cases, the invoices that didn’t go straight through. And sometimes you don’t want straight-through processing; sometimes it’s an invoice you don’t want to pay. So complete touchless processing isn’t the goal — maximizing throughput of invoices you do want to pay is the goal. The way we use AI is to detect patterns in user processing and then propose rules that our automation follows. Every time an invoice goes through the cycle and gets paid, you can see exactly which rules that invoice followed and why. There’s no black box. No ‘the AI just picked that.’ Everything is auditable and traceable back to rules and who approved them.
AI has really changed the way software is being made — it’s much easier to develop software today. But that doesn’t remove the engineering that needs to happen behind the scenes to make things work well. We have a great engineering team behind Yoga. We’ve built it cloud-native from the ground up so it can scale to any invoice volume. Yoga currently processes about five million invoices a year across all our clients, and it does it performantly — and we can continue to scale well beyond that.
This is fully hosted and managed by us. We’re constantly tuning the automation behind the scenes to maximize efficacy as vendor formats change, as your vendor population changes, as you acquire new organizations and new vendors come in. Our team and our tools are constantly monitoring and acting on opportunities to improve automation. We also bake in automated modifications as part of the service — if something about your business changes, you need to capture new fields, or you need to change how a workflow works, all of that is included. We call this a solution as a service rather than software as a service. We have a team of Infor and IOFM-certified consultants available to support you. Your main implementer stays with you as your permanent liaison, but they have a team behind them to make sure you’re always covered. Your KPIs are our KPIs — we’ll show some reports that cover the AP operations metrics leading organizations track: days payable outstanding, how long it takes from invoice receipt to payment, and so on. Reducing that time to your goals is our goal as well.
The pain points Yoga is built to tackle: real-time visibility into your liabilities, pre-close accrual confidence, improving coordination between AP and purchasing on exception invoices, making sure every invoice is auditable, and avoiding fraud — increasingly important in today’s generative AI era. Confidence in accuracy is part of our service. We make sure your systems and processes are aligned so you have the right approvals set up, there are no gaps in how approvals are defined, and there’s no hot potato situation with an invoice bouncing around unanswered.
We also have a dashboard that lets you optimize for either maximizing early payment discount capture or managing cash on hand — showing invoices in those different lights. And one thing Yoga handles well is multi-business unit complexity. Large organizations with several business units can manage that easily in Yoga.
Let’s jump into the demo. There are several ways invoices can get into the system: you can capture them directly from disk, email them in, pull them from an API, pull them from an SFTP, or upload an Excel sheet of invoice lines and we will generate invoice images from that sheet on the fly. Whatever way you receive invoices, we can bring them in quickly.
I’m going to start by capturing some sample invoices here. Capturing from disk is probably not the most common approach — the most common way is that we monitor your email inbox and poll it automatically, and I’ll show that in a moment. But when you capture from disk, you can review all the invoices before adding them to make sure you didn’t accidentally upload a resume or a bank statement.
Within just a couple of seconds of adding those invoices, you can already see all the extracted data. You know the vendors, the invoice types, the liability amounts — all right at the edge of where you receive invoices. We run all the typical validation you’d do before posting an invoice into the system, and we highlight anything that’s missing. If an invoice could have gone straight through, you can turn that on or off. A lot of clients like to turn it off initially just to verify there are no false positives in the straight-through logic before enabling it. I’m going to go ahead and delete these, though, because the way I really want to bring invoices in is via email.
To run the email job — which in production would just run automatically every minute or every five minutes depending on what you work out with your IT team — we run our email connect job. All our invoices are now in the system from email. The ones that could have gone straight through: we pulled all the information off of them and automatically populated the routing category based on learned rules. As users process invoices and change field values, the system records every change in the background — what came in from OCR, what the user changed, and what it was changed to. Once enough changes accumulate, the system recognizes the pattern and suggests a rule. That rule is back-tested for efficacy. If it meets the efficacy threshold, it automatically approves itself — or you can configure it so that rules require manual approval. The system is constantly learning from user feedback.
If an invoice comes in via email, we place the email itself in Supporting Documentation. So if there’s important context in the email body — a department reference, a missing PO number, anything like that — you have both the primary invoice document and the supporting documentation together. You can also add additional supporting documentation manually: email chains, attachments, whatever is relevant to why an invoice was paid a certain way.
Because all the fields are filled in on this one, the Post to Infor button is available. I’ll add a comment — ‘Hello webinar’ — and post it. It takes me right back to my queue so I can keep working. You can also configure the system to automatically load the next document based on your active filter. If you’re working a particular vendor’s invoices and have the list filtered for that vendor, as you post or route invoices, the next one for that vendor loads automatically so you can crank through them quickly.
Now let me quickly show what this posting looks like end-to-end. Here’s our invoice in Awaiting Payment. I’m in Infor CloudSuite now and I’ll search by invoice number. You can see it came right into Unsubmitted. I have automatic approval submission turned off right now so I can walk through this — but in production, if the routing category is populated, the invoice would automatically go out for approval as soon as it’s posted, so your approvers would get it with the routing category already in place.
Opening the invoice in Infor: everything came over. In the Documentation tab, all the notes from Yoga carry over as well. If there was back-and-forth commentary in Yoga about why something was set up a certain way, all those comments come into Infor along with the invoice image and supporting documentation. You’ll see them in the Related Information widget — the invoice image and supporting documentation open directly in the browser as PDFs. Once the invoice is out of Yoga and in Infor, Infor should be your system of record for document images and processing history. The invoice will go out for approval from here. Same for PO invoices: when we post them, we immediately attempt to match. If the receipt exists and the invoice matches the PO, it will automatically match and get released for payment.
If there are matching issues — the receipt doesn’t exist, or something similar — the invoice will sit in Unsubmitted until your MA126 runs and picks those up. Any buyer messages are resolved through the standard Infor process from there.
Going back to the AP Review queue: I mentioned that the system can automatically learn rules, but you can also manually set them up. Sometimes the conditions for a rule are complex enough that you want to define them yourself. For this routing category, we didn’t auto-populate it because we didn’t have enough samples yet. So I’m going to set up a rule manually for this vendor.
In Intelligent Vendor Management, you get a full dashboard for each vendor: spend, invoice types, reason codes for why invoices are stopping, which fields users are having to change, and what the straight-through processing rate is. You can drill into every vendor individually. On the Field Rules tab for the invoice routing category, I have a pending rule I pre-set up but haven’t approved yet. Let me build one from scratch so you can see how it works. We have Declined as a separate status from Deleted — if the system suggests a rule that doesn’t make sense and you decline it, it stays in the Declined list. The next time the system tries to learn that same rule, it checks the declined list and won’t re-suggest it. I’ll add a new rule: for this vendor, every time the customer ID is DMC128, set the invoice routing category to Financial Accounting. I’ll save and approve that rule.
Now let me recapture the Sunline Catering invoice and reprocess. A moment later, Sunline Catering is now straight-through eligible because all the required fields are populated — the routing category was set by the rule we just created. That’s the manual rule approach. The system also picks up these patterns through processing automatically, so you have flexibility for both.
Let me briefly cover the document notes. Say we have a PO invoice with a line total mismatch — we were only able to match one line, and three others are missing from the PO. I’ll flag this as Missing Line and route it to the Purchasing Invoice Resolution queue. Whenever you route an invoice to another queue, the system requires a comment, because you’re handing this off to someone else and they may need context beyond the error message itself. I’ll add: ‘Researched this invoice — there are missing PO lines. Please advise.’ That routes to the Purchasing Invoice Resolution queue. The buyer team comes in here and sees the last comment right on the list view — it’s immediately clear why something is in this queue. We also surface who the buyer is on the purchase order, so the list can be automatically filtered to show each buyer only the invoices assigned to them.
In this queue, buyers can’t edit any invoice data — the form is locked. They can only add document notes. They look at the issue, go add the PO lines or advise to short pay, and then route it back to AP Review with a comment: ‘Added lines to purchase order.’ When it comes back to AP Review, it carries a reason code of Returned from Purchasing Invoice Resolution. You can quickly filter by reason code to see exactly which invoices have come back from purchasing and are ready for AP to finalize. You can also see the buyer’s last comment right there. This takes conversations that today might live in email and puts them all in one place around the invoice — and all those comments get recorded back into Infor.
Jumping ahead to reports — specifically around purchasing invoice resolution: we track all the events between AP and purchasing. We know how often invoices are bouncing back and forth, how long things have been sitting in the purchasing queue, what the dollar value of held invoices is, and who’s processing what out of that queue. As an AP manager or AP leader, you have full visibility into where invoices are and what’s happening to them.
Our PO line matching uses a custom algorithm that prioritizes description, then vendor item number, then Infor item number to match invoice lines to PO lines. From there we look at unit price and unit of measure. We support substitute units of measure and line-level tolerance thresholds. The matching logic is sophisticated.
A few other reports and dashboards: document ingestion breakdown by type — PO versus non-PO, email versus EDI. Processing differential — are you processing more or fewer invoices per day than are coming in, so you know whether you’re staying ahead of or falling behind on volume. A queue snapshot you can drill into. User productivity: we show both Post to Infor route actions and total route actions — so you see not only which invoices are going to payment but also what’s being routed to cancel, purchasing, or duplicate. Yoga is typically the most productive user in the system from a straight-through processing standpoint.
Taking a live question here — Danielle asks: the matching is done in Yoga; if so, how is the information communicated back into Lawson? In Yoga, we only match the purchase order line to the invoice line. That match selection is communicated back to Infor via the ION API, which creates the invoice and the line match directly in Infor. From there, if the receipt already exists, we run the attempt-to-match process — essentially the same as manually hitting the Match button before releasing an invoice — and if everything checks out, the invoice gets released. If the receipt is missing or there’s an out-of-tolerance line amount discrepancy, it creates a buyer message just like it does today when you enter an invoice manually.
My favorite report is the Monthly Metrics Report — our shared KPI report that we work to continuously improve together. This is a snapshot of a live client’s data in our demo environment, showing what a fully mature version of this looks like. Month by month: invoices paid, checks cut, PO versus non-PO percentages, which came through email versus EDI, average invoice date to check date, average time from invoice date to creation in Infor — broken down by PO and non-PO, and whether available discounts are being captured. It also shows the average time invoices spend in Yoga before they go into Infor — that’s the window where Yoga is processing them, and you want it to be minimal.
For this live client, it’s been mostly one to two days, with a few spikes to three or four days in recent months. They predominantly process PO invoices through Yoga. They’ve been sitting above 70% straight-through processing on PO invoices for the past five or six months. When they acquired a new hospital last November and brought in a large batch of new vendors, the straight-through rate temporarily dipped — but it came back up within a month.
We also have an accrual report that you can export and upload into Lawson, showing all invoices by vendor and amount that are still outstanding for accrual purposes. And our cash flow and discount optimization dashboard gives you a snapshot of outstanding invoices, average payment time, and invoices paid in the last 30 days. On the cash flow side, it shows what’s due when across your Yoga queues. On the discount maximization side, it shows which invoices have discounts expiring within specific time buckets so you can prioritize accordingly. All of this is sourced from our Databricks data warehouse. We provide these out-of-the-box reports, but we can also help you build reports in your own BI or warehousing tools using this data.
A couple more features worth calling out. The invoice image and form can be popped out onto separate monitors and will stay in sync as you move through invoices. All views are customizable — columns, sizing, positioning — and those preferences persist per user. The document notes panel lets you filter by comments, by form data changes, or by full history. Every time a field is changed, we record what it was changed from, what it was changed to, and who changed it — fully auditable.
You can split and merge documents directly from the invoice view. If a single PDF contains multiple invoices, the system classifies and splits them automatically on ingestion. If that doesn’t work for some reason, you can split manually using our split UI. You can also merge documents, download documents, download the OCR results, download the form change history, and send an email directly to the vendor from within Yoga. The vendor’s email auto-populates, you type your message, attach the PDF, and it goes. We can configure this email to come from a shared AP customer service inbox so that vendor replies come back into that inbox and CC the original sender. You can also add your own custom signature so the email looks like it’s coming from you rather than a system. This gets used not just for vendor communication but also for emailing requesters to ask about invoice approvals or coding before an invoice goes out for formal approval.
For invoice capture: we have a basic check request form built in out of the box. We’ve seen this used for non-PO check requests and for employee reimbursements. The Excel invoice ingestion — uploading an Excel sheet and generating invoice images from it — is also available. The Vendor Rules Command Center shows all the rules the system has learned, any pending rules waiting for approval, each rule’s efficacy, and the ability to approve or decline them. You can view this as a feed across all vendors or drill into individual vendors in the Intelligent Vendor Management section.
That’s the bulk of what I wanted to show you today. The one thing I only have a screenshot of right now is our Statement Matching solution. As statements come in from your inbox, they go through a special workflow where they get OCR’d, and the system attempts to automatically match statement lines to invoice lines already in your system. It allows you to email vendors for missing lines and track accruals for invoices you may not have received yet — giving you a more accurate cash forecast. As those invoices come in, the system will alert you to accruals that should be reversed. This is in beta and coming out in Q3, and it solves the other side of invoice processing: once you’re through the day-to-day grind, you still want to verify that everything the vendor is billing you for is accounted for. We’ll be doing a separate webinar on this one shortly.
