In 1990, ASF ran a number of advertisements with this one bold statement: “We don’t make a lot of the products you buy. We make a lot of the products you buy better.” We may be having one of those déjà vu moments now as a number of AI/ML and RPA firms are doing something similar with Finance, HR and ERP applications. These new vendors aren’t building more general ledgers and payroll applications but they are making it possible to get piles more value out of those old-school, transaction-processing application software products.
While technologies like AI/ML and RPA have been around for a while, the power and ease of use for these has improved dramatically in recent time. It’s now common to find solutions built on modern platforms. The new tech stack within these means that users can:
- Develop new white space apps with no- or low-code tools.
- Develop these apps without IT help.
- Integrate a growing number and type of databases to these tools.
It’s as if these tools are going through a democratization phase. What this implies is that users (not just IT) can build their own smart apps (ie, not wait for vendors to create new apps for them), build them quickly, and adjust the way these apps work based on evolving organizational and other requirements.
With these tools, users can:
- Create exception handling utilities that route specific and non-standard transactions to specialists for additional review or handling.
- Identify potentially fraudulent transactions.
- Spot potential bribery events (ie, violations of the Foreign Corrupt Practices Act) BEFORE they happen.
- Monitor the real-time impact of social sentiment on top-line revenue, future hotel bookings, restaurant reservations, etc.
- Assess the impact of future weather situations on staffing needs, sales, etc.
- And more.
But, before we get caught up in the recent hype, let’s get grounded first. Solutions that purport to automate processes would more accurately be described as automating a task (eg, determine which expense account should be debited for a specific invoice), function point or decision than an entire process (eg, order to cash). These tools can shave days/weeks out of approval timeframes. They can rescue accounting, HR and other users from the tedium of dealing with repetitive events (eg, processing employee paid time off requests). And possibly one of the more important benefits comes from identifying new, non-standard or unfamiliar events that warrant human scrutiny or concern.
The bottom line is that this new generation of technologies can potentially:
- Refocus work on the things that really matter.
- Eliminate a lot of non-/low-value-added work.
- Shorten the elapsed time of processes.
- Improve a function’s (eg, corporate accounting) service levels to its constituents.
Where’s the market for this?
So, you’re probably thinking “We’ve already automated our financials. In fact, we’ve re-implemented them 3-4 times since the 1970s. We’ve already gotten all of the transaction processing efficiency out of these apps. Is there really anything left?” The short answer is “Yes”.
What a lot of ERP, Finance and HR apps did was mimic old manual processes. If the old process required every invoice to go through the same approval process, be entered (by a human being) into the application, follow a number of approval and validation steps, etc. then your automated solution probably mimicked that. Did the elapsed time to process transactions actually go down? If your firm is like many others, it didn’t. Aspects of the old process are done more efficiently but the net effect of it all is that it still takes your firm a lot of people and time to close the books, pay bills, respond to employee and supplier inquiries, etc.
Let’s face it, the application software you have may not have been designed to:
- Process individual transactions (eg, invoice approvals) via different processes based on unique aspects of the transaction (eg, total dollar amount of invoice) and who must handle it.
- Assign a risk profile to each and every transaction and then route them to approvers based on the transaction’s risk profile.
- Serve constituents beyond the processes’ usual/frequent users.
- Marry external big data sources to transaction data to assess the likelihood of fraud being present.
- Detect patterns in transactions that might indicate potential criminal activity.
Corporate ERP users are finding lots and lots of ‘opportunities’ to plug value leakage, service failures and inefficiencies. In January, I documented how Automation Anywhere’s CHRO and 30 of her HR staff built an array of bots to handle all sorts of issues, some of which were new due to the pandemic. You should check out this recent video by Slack showing how they used Workato technology to develop process automations in numerous functional departments (note: Be sure to check out the graphical list of automations near the end of the presentation). That video is also important as it highlights how collaboration and RPA bot technology can co-exist well and deliver value.
Let’s dig deeper in several vendors’ modern approaches…
Workato and the rise of advanced technologies
Workato is a process automation vendor that has been growing rapidly. It has a number of IT, Finance and HR customers using its tools. I recently spoke with their CIO, Carter Busse. His firm had recently completed a study on the uptake of these new technologies. The report indicated:
- Workato’s data shows that one-third of enterprises were using automated processes in five or more departments during the past year, up from just 15% in the year before the pandemic began.
- Customer Support automations saw the biggest growth of any department, up by more than 290% year-over-year.
- Automation is a team sport between business and IT: IT users accounted for 55% of automations created, while business users accounted for 45%.
- Collaboration apps like Slack and Microsoft Teams were used in 20% of all automated tasks.
- Recruitment saw the highest automation growth of any single process, at 547%.
- Finance automation is a growing priority, with the volume of automated Finance & Accounting processes increasing by 199%.
Carter indicated that HR and other executives’ awareness has grown a lot in the last year or so. Factors like COVID, work from home, new business models, etc. have clearly pushed functional leaders to consider new solutions, especially solutions that can be implemented quickly, at low cost, with the ability to be easily modified and/or to integrate with existing applications. He added that buyers are more and more aware of potential benefits especially in ways these technologies can provide "competitive gains and (increased) efficiency." I’d concur.
When you look at the numbers above, some items really jump out. Recruiting is clearly an HR trouble spot today if only because of factors like the Great Resignation. CHROs are struggling to fill open spots and they need tools to automate low or no value-added tasks (like scheduling interviews). Beyond recruiting, there are a number of other HR opportunity areas ripe for RPA/AI/ML exploitation like onboarding, paid time off requests, benefits enrollment, automated bots to deal with common retirement planning questions, etc.
At Workato’s virtual user conference last week, the company also introduced their new Back-to-Office Accelerator. This pre-packaged product:
… contains everything users need to automate back-to-office processes, all easily orchestrated by Workato's platform chatbot. Businesses can use the package as-is, or customize to their specific needs and get going in a matter of days. These automations help facilitate a safe return to the office by allowing employees to do the following via a bot on Slack or Microsoft Teams:
- Prove they are vaccinated.
- Reserve a desk space in the office.
- See who else is coming to the office on any given day.
- Follow their manager or teammates and get notified when they plan to go in.
- Contact tracing.
A different data point from the CPM/EPM front
I heard (enterprise performance management vendor) OneStream CEO Tom Shea speak at their virtual user conference (Splash 2021) this week. His talk was probably one of the clearest and most coherent descriptions of how apps are evolving, especially how advanced technologies are maturing. Tom described the different generations of software and the implications of each generation’s design on delivering insights and value. The remarks got really interesting as he started to address how advanced technologies like AI/ML will impact companies in general and how these capabilities will impact software like the kind that OneStream sells.
Some of Shea’s key points included:
- Large scale adoption of early ML/AI tools was hampered by a severe lack of data scientists. The earlier tools required people with significant technical skills to create and adjust these tools. The result was that only large firms could afford the technology.
- Earlier solutions were often fragmented, lacked transparency, and didn’t scale well.
What the market needed was:
- Vendors to make a number of pre-designed solutions that end-users (not IT) could deploy and tune. These solutions would not necessarily require data scientists.
- Two different kinds of solutions were needed. One that focuses on internal data to aid in predictive analytics and another set of solutions that utilize a variety of internal and external data sources to find otherwise hidden correlations.
There are definitely concerns with either kind of solution. The predictive apps, because of their reliance on prior, limited, internal transaction data can miss emerging trends. If your firm experiences a seasonal jump in sales every Q4, the predictive tool will likely reflect that even though external data might suggest a less optimistic forecast. In the case of flight risk (attrition prediction) software, tools that rely on internal transaction data might miss scores of other predictive factors from external data sources that could support or dispute the risk prediction from such limited datasets.
The ML/AI apps may use databases that may not be free from bias. If so, the tools will likely replicate the same bias in the results they present to you. For example, if a sales database lacks records from women buyers, then the recommendations/insights it generates would likely be irrelevant in understanding women buyer preferences.
What Shea sees are customers who need solutions that:
- Have some initial capabilities pre-built but can be tuned to a customer’s particular vertical or product/service lines.
- Allow users to create potentially hundreds or thousands of different models based on the use of different internal and external data sources.
OneStream previewed a series of solution capabilities to address modern business needs. Some capabilities include the ability to create thousands of ML models in parallel, reduce a user’s dependency on data scientists, and make AI/ML tools explainable to managers, auditors or regulators, etc.
More specifically, OneStream stated in a press release this week that its:
…new AI services and ML capabilities will allow finance and operations teams to easily incorporate advanced forecasting and other ML techniques into their existing planning processes. Key benefits of the solution include:
- Analyzes vast amounts of data generated from internal sources such as ERP, CRM, and Supply Chain systems as well as external sources such as weather and other macro-economic data, building thousands of ML models in parallel.
- Delivers AI results and ML predictions that are seamlessly and transparently consumable in OneStream analytic models, giving businesses the efficiency and confidence required to derive the most value out of advanced ML predictions.
- Solves common business problems such as demand planning, revenue forecasting, anomaly detection and cohort analysis.
- Supercharges productivity for in-house data science resources by proactively delivering thousands of forecasts in a fraction of the time compared to traditional ML modeling techniques or other solutions on the market.
- Supports effective eXtended Planning and Analysis (XP&A) across the enterprise.
It’s interesting to see how a new vendor sees this changing landscape. For that, Auditoria aims to plug productivity and insight gaps in the back office. Their focus can also provide valuable insights into what shared services organizations should be considering.
One exhibit of Auditoria’s shows a number of ‘skills’ or new autonomous applications available from their SmartFlow platform (see below).
Auditoria has spent time learning the ‘taxonomy’ of leading ERP solutions. Why? This helps customers quickly connect Auditoria to their ERPs and it speeds time to value. It sees its bots as being ‘pretrained on process hygiene’. I like the concept and customers definitely want apps not just a pile of tools or a bare-bones platform. One focus area, according to Auditoria founder Rohit Gupta is "cash performance as an asset." This looks across a company’s value chain to find ways to improve forecasts, reduce losses/fraud, etc. So, like the BASF anecdote at the top of the piece, Auditoria isn’t going to make the old-school apps found in ERP software. They’ll make those old apps more valuable/better with Auditoria’s new apps – apps that add value today.
I have yet to meet a CEO, CFO or CIO who wants to upgrade or replace their ERP software unless there’s simply no choice. Why the reluctance? An ERP replacement project:
- Is often risky.
- Often creates processes and outcomes little changed from the current system.
- Can be very costly and end up costing much more than the company, integrator and/or vendor estimated.
- Can disrupt current operations, customers, orders, etc.
- May not deliver benefits for some time.
- Is often very time consuming.
Newer solutions, ones that complement (not replace) the existing ERP, may be a better alternative as they:
- Go in quickly.
- Don’t require the huge effort that an ERP replacement would require.
- May be lower risk to implement.
- Deliver very different kinds of net-new value.
These new tools/apps have a new vernacular: autonomous, intelligent, self-driving, low-code … Some of that may be marketing hyperbole while much of it may be fact. You’ll need to judge each accordingly.
But beyond the coolness of the new tech involved, there’s a very practical or pragmatic reason we all need to pay attention to these new solutions. The world realities now are much more volatile. Businesses cannot afford the cost, time and risk to re-implement or change ERPs with every new regulation, new business model, M&A event, pandemic triggered change, new generation of worker, improvements in consumer tech, etc.
The world got more volatile and businesses need tech to be more dynamic. That’s the new world and old ERP has a role in it albeit a different one than before. Old ERP will work in the new world but it needs to be supplemented with powerful new capabilities that add current, relevant value. It’s time to re-think ERP.