I try not to develop a rooting interest in a software program. That's not why I'm in this game. But it's hard not to root for SAP Business ByDesign, a perennial underdog that absolutely refuses to die.
No, SAP didn't always seem to care much about ByDesign, but its customers always did (During a recent podcast discussion on transformation and hype with Lucy Thorpe, I elaborated on ByDesign's extraordinary fortitude).
Today, Business ByDesign is no longer on the chopping block, but there are new issues to consider.
Some pundits assume ByDesign is withering in the shadow of SAP's S/4HANA pursuits, but there is an intriguing flip side to that story. The company-wide re-platforming efforts around the SAP Business Technology Platform (BTP) means that the SAP ByDesign team is now able to share resources with the S/4HANA team, including AI models and predictive libraries.
Business ByDesign's "intelligent automation" bot rollout
That raises a different question: is it possible for ByDesign to offer the same type of "Intelligent Enterprise" roadmap SAP offers to S/4HANA customers? The SAP Business ByDesign team believes that answer is yes - this ByD Intelligent Enterprise video explains why.
To frame this challenge, let's limit the definition of an "intelligent enterprise" to: effective use of automation and machine learning to improve operations, decision making, and customer relationships. One thing we know for sure. If small and midsize businesses are going to achieve this, they need help - as in, packaged "intelligent" applications and embedded AI. Few businesses of this size will be able to afford data science teams. They expect help importing data and tuning the results. A platform to build or extend "AI apps" is not enough for this market.
So how is Business ByDesign accomplishing this? This SAP-authored blog post on best practices in intelligent automation for ByDesign gives the update. Short version: SAP is approaching "intelligent ByD" in phases. The first phase is what SAP calls "intelligent automation." In brief, the roadmap is:
- Intelligent automation - already in production.
- Decision assistance and decision making with Hana predictive analytics
- "Next generation user experience."
As of the August 2021 ByDesign release (2108), customers had access to six "best practice bot templates" via the SAP IRPA Store. SAP's Simona Marincei, Head of Intelligent Processes – VP, Product Management SME, who authored this post, says these bots are "available out of box and ready to implement in minutes."
These bots address common workflow automation issues, for example:
- Business Document Extraction from E-Mail
- Supplier Invoice Upload for Intelligent Invoice Scanning
- Sales Order Creation from Customer’s Purchase Order
Customers can customize the bots, or, alternately, design their own. Though the ByDesign bots are fairly new, the adoption stats are encouraging. During our July interview, Marincei told me:
Since launching invoice scanning, the first such service in August last year, up to now we have 260 customers from twenty countries in six languages. Out of these, 106 are already productive. We have over 160,000 invoices scanned in seven months.
The increase is outstanding... Every single month is growing the amount of documents that are scanned using these services. Also, we also have quite a healthy adoption pipeline, with 300 tenants that are provisioned, but not yet used. These are customers that requested to have the service, but they didn't get started yet. So we are expecting tremendous success, especially that they come at no extra cost for ByDesign customers. (Author's note: though initial bot access is free to ByDesign customers, there are licensing implications for customers to understand that are beyond my scope here. See more details on bot pricing here).
Making intelligent RPA practical - a ByDesign scenario
So that's ByDesign's strategy: make "intelligent automation" practical, embed it into existing apps, and make it free/affordable. What informed these product decisions? During summer briefings with Rainer Zinow, SVP in SAP's Business One & SAP Business ByDesign group, and various members of his ByDesign team, I dug in.
In a ByD AI discussion, Marincei cited the hypothetical example of Mary's bakery, "buckling under the enormous weight of paperwork" required to keep her business up to date. Enter "intelligent document processing." SAP's Intelligent Enterprise for SMBs video uses a similar customer example. I don't know about you, but I don't need much persuasion that smaller businesses can use all the document automation they can get. But how can machine learning and AI play a role?
In other words, how do we move from RPA to something we can legitimately call "intelligent RPA," and not just because the marketing team wants to update their slide decks? Marincei spoke to several aspects of this, starting with invoice processing:
Instead of manually downloading, classifying and saving her supplier invoices from emails and then typing them one by one into SAP, the intelligent invoice scanning, hand in hand with an automation bot, will do that for her - using a mix of SAP AI business services and SAP Robotic Process Automation integrated in, or working side by side with her SAP ERP system.
Next, Marincei touched on receipt and purchase order processing:
What about all those receipts she and her sales team got when they joined the Food Fair last week, which now need to be added to her travel expense report? The intelligent mobile scanning will do that for her using a mix of OCR with machine learning models, embedded in her SAP Business ByDesign mobile app. Mary used to receive a lot of purchase orders via emails from event organizers. Now instead of her manually creating sales orders by copying and pasting the relevant data from the email purchase order, the intelligent RPA bot will take care of that at the scheduled time, extracting all the necessary information with the help of the document information extraction service, and creating a draft sales order for her and SAP Business ByDesign or SAP Business One. Today, all Mary needs to do is validate the drafts and deal with the exceptions.
Then the punchline:
Mary is more than happy with the intelligent business document processing capabilities, because now she can focus completely on the aspects of her business that really matter to her.
If we can achieve that, we'll hear no arguments. This is why we should prize so-called "business outcomes" over quarreling about whether this is really AI, or how "intelligent" it actually is.
The Business ByDesign AI roadmap
And what about the next two phases in the AI-for-ByD roadmap? Marincei spoke to decision support and approvals.
Back to our bakery example: if a customer has a purchase order of 100 kilos of flour, based on the previous data, ByDesign's ML algorithm can calculate the prediction. If you are within an expected range, that order could be automatically approved. But if it exceeded those limits, an approval would be surfaced for the manager to check. Thus you have a mix between traditional workflow approvals, and what SAP calls "advanced intelligent technology." Though as Marincei points out:
For the end user, it does not really matter what kind of technology, as long as you are solving an end-to-end business problem.
As for next-gen user experience, that's all about using chatbot interaction to guide you into today's priorities, rather than combing through screens. So instead of navigating documents, you would ask the bot for today's agenda, and be directed towards your action items, e.g. open orders to approve. Or you could ask the chat bot to create the relevant invoice for a delivered product. This sounds straightforward, but I don't believe it's easy to achieve in an intuitive way. I'll be interested to see how SAP fares.
My take - what is the ROI of practical AI?
This roadmap reflects the more fruitful approach of being ruthlessly practical with AI: making existing workflows easier, and using AI's pattern recognition strengths to surface problems or action items. Often called "augmented AI," if this is done at a role-based level, I believe it can have business impact.
The SAP Business ByDesign team estimates that the average added value generated by their "first wave" of intelligent RPA ranges from £47K to £242K. During the aforementioned video, Alan White, VP of Information Technology with Business ByDesign customer SEF Energy, spoke to the pain of processing 13,000 to 15,000 manual invoices a year, roughly 10 to 15 invoices per hour. Did using intelligent invoice processing help? White says:
Using the new feature, we only need to validate and eventually correct the scanned document before submission, which is really reducing the manual processing time to 10 to 20% of what we needed before.
And the financial results? White:
So if we take these numbers and apply them to the fully loaded cost of an employee per year, we end up with an annual savings somewhere in the region of $27,000 US dollars, and that's really a conservative use case. In an optimistic scenario, it's probably closer to $70,000 US dollars.
It's amazing what a difference platform unification makes. Historically, SAP Business ByDesign had to make do on the margins with resources. Today, they are taking advantage of HANA's predictive libraries, now built into ByDesign, and: they can utilize S/4HANA's machine learning investments. As Zinow wrote to me:
Simona works hard to deliver on the promise of AI (everything around business document recognition), predictive analytics (supported decision making) and automation (hands-free ERP, automate whatever you can). The reason why we can do this is that we can get that from the corresponding SAP AI/ML etc. teams. And they are happy to have such a passionate 'customer' in Simona.
Does that mean that these services are truly "intelligent"? We can have a good debate there, as I did on the ByDesign podcast. Intelligent, as I define it, means the system is truly learning, as its database of interactions grows. We didn't have time to sort that during our ByDesign AI briefing, but as I've noted, I think that's something of a semantic debate - what's more important is perceived customer value. So far, based on the testimonials and adoption data SAP has provided, that looks promising, though I'd like to hash this out with a customer or two in more detail.
Whenever I bring up SAP Business ByDesign, readers want to debate its future, but I think that's getting old. Yes. SAP's massive S/4HANA investments do raise questions about where its non-S/4 ERP investments stand, but I believe as long as Oracle has NetSuite, SAP will now have ByDesign. ByDesign's ability to take advantage of AI/ML and platform resources bodes well. SAP Business One is another story, and a debate for another time.
I do think SAP should make more PR headroom for ByDesign. Yes, Sapphire Now will always be dominated by S/4HANA, but I don't see any reason why Business ByDesign shouldn't have a dedicated user conference, a la SuiteWorld. It's hard to take ByDesign as seriously as some of its competitors without that type of showcase event. Yes, to some extent, such an event is symbolic, especially in our hybrid event world, but it's also a visible vote of confidence - and a major step forward in building a customer community.
Also, and I know SAP absolutely loves when I bring this up, but as a multi-tenant advocate, I still view ByDesign as the SAP ERP product with the clearest, most open, and most historically-rooted commitment to multi-tenancy. I wish I heard more of that talk from the S/4HANA cloud team, though I understand why they don't lead with that these days. I do like the emphasis on moving from custom to standard on S/4HANA that practically all SAP executives, from CEO Christian Klein on down, has been preaching. But there is a reason why the most successful SaaS software is predominately multi-tenant, or, if you want to get semantical, multi-tenant or multi-instance.
Bottom line: SAP Business ByDesign may not grab the headlines, but it has some important things to offer, both to customers - and to SAP's own ambitions to be a modern (cloud) ERP leader.
Updated, 7am UK time on Saturday, October 2nd, with a number of small tweaks for reading clarity.