Autonomous ERP

Imagine an ERP solution that drastically reduces manual entry by 90%! The remaining exception transactions can be created via Natural Language Processing (NLP) or user confirmation of AI/Machine Language (AI/ML) recommendations. All supplier invoices are electronic. Moreover, bank reconciliations are performed near real-time via blockchain. Supply chains have the ability and capacity to process an order request, gather the commodity from the location, and deliver the unit to the desired delivery point without human intervention1. Further, sub-ledger reconciliations are eliminated because there is no more need for sub-ledgers. Reports become an obsolete way to communicate information. ERP Cloud updates, testing and feature deployments occur without heavy reliance on third-party consultants.  If interested, please allow me to share my vision of autonomous ERP. 

What is Autonomous ERP?

While there is no agreed-upon standard definition, following is my characterization of autonomous ERP is based on research and my ERP delivery experience.

Key Points

  • Reducing and/or eliminating data entry is strategic for autonomous ERP. Robotic process automation (RPA) is a short-term fix to automate data entry, but the goal is to eliminate the need for data entry altogether! IoTs must expand and interconnect, facilitating machine-to-machine communications. Sensors become self-sustaining 2
  • It is a reality that ERP Cloud is the delivery model of the future, since its economic and technology-scalable advantages surpass those of any other ERP delivery models. However, if you do not trust your ERP Cloud service, you will not be able to evolve with it. The trustworthiness of ERP Cloud services is only proven through repeatability, reliability, and capability in achieving desired business results. Merely having “leading” features and technologies does not an autonomous ERP solution make.
  • Just as OLAP and HTAP/In-Memory DBMSs are a necessity for EPM performance today, these are in fact not a requirement in a solution whose performance is exponential compared to today’s standards.
  • Many existing constructs and ERP designs are still based on the current limitations of manual effort and technology. As Brian Sommer stated:

“Businesses have to be ready to move away from monthly, aggregated data to more precise data, costs and insights.”7

Trust and the ability to rollback AI/ML-initiated transactions will be the key drivers in ERP autonomy adoption. 

Five Levels of ERP Autonomy

Since there is no standard definition for ERP autonomy, there is no clear definition for its different maturity levels either. I submit the following model for your consideration:

Key Points

  • This model is based upon the Department of Transportation (DoT) definition model for autonomous driving.
  • Level 1 represents siloed feature ML capabilities that represent a “proof of concept” approach from ERP Cloud vendors.
  • Business results are created through business processes, not individual functional activities or features. Given this reality, a vendor who can provide autonomy across a business process is considered a maturity improvement. (Level 2)
  • Quick adoption is possible only with a rollback feature. This capability will enhance user adoption and reliance. At this point, over 80% of transactions are digital (invoices, payments, customer interaction) (Level 3)
  • As the ERP solution learns (digital experience) and business users trust the ERP AI/ML recommendations and results, customers will have their organizations focus less on transaction processing and more on decision-making activities on a trial basis. (Level 4)
  • At level 5, Business users “think outside the box” to ensure customer value and the ERP Cloud service performs the necessary tasks.  The human brain, with its superior neural network, is fully leveraged.

Why do we need Autonomous ERP?

Given Moore’s law on technology innovation, I anticipate that autonomous ERP will happen within my lifetime. Autonomous ERP frees the human mind from the tedious and mundane. As millennials and Gen Z-ers expand their numbers and influence in the global workforce, there will be a greater demand for human creativity within every job role. Today, human creativity is limited by technology performance barriers which require humans to support business transaction processing.  However, small strides are being made.  Consider the following advancements:

  • Although separate data models (e.g., OLAP) are a requirement for EPM performance, they are not a requirement in a solution whose performance is exponential according to today’s standards. Making ERP data and analytic structures “in memory” is a step in the right direction. However, the effort required for autonomous ERP will require more than shrinking the von Neumann model.
  • Just as compiled code was a necessity for application performance, so is the sub-ledger to the accounting model. The sub-ledger is a necessary construct based upon established constraints and limitations.

From my perspective, I see technology as finally catching up with business and not technology changing business.  Technology greatly enables the human brain to think and express thoughts outside physical boundaries whereas AI/ML and Deep Learning are limited by the data captured. In fact, human thought is not bound by such a constraint and is, in my humble opinion, the most under-utilized resource with the greatest potential for creating business value.  Consider the following illustration below regarding a specific use case for autonomous ERP.

Let’s look at the example of a day in the autonomous ERP world. Jill is the controller for a mid-sized company. As Jill prepares for work, she gets the “morning report” with the following:

  • Tasks to perform that day as well as tasks that are past due.
  • Overall performance of her organization versus the plan.
  • Business exceptions that must be addressed manually.
  • Recommendations to address current business activities and exceptions.
  • The “morning report” is available via all interaction channels (UI, Voice, UI plus Voice, etc.)

Subsequently, Jill provides verbal commands based upon the morning report.  The autonomous ERP Cloud service executes transactions based upon Jill’s decisions. Next, Jill focuses on the continuous accounting business process to review scheduled activities and provide guidance as needed. The third step of Jill’s day is to review recommendations and approve the exceptions to the standard business process. Now with the basics addressed, Jill can focus on improving the existing business process and prepare for emerging demands. Scenario analysis with business impact scoring and Jill’s input (most important) enables autonomous ERP to make better decisions. Finally, the autonomous ERP solution increases Jill’s productivity by implementing “quick win” capabilities and guides Jill and her team through the deployment.

Customer Recommendations

Autonomous ERP, Intelligence ERP or “plug-in-your-name-here” ERP sounds tempting. However, there are four areas that you should investigate and evaluate for an ERP vendor’s strategy regarding autonomous ERP.

Key Points

  • If an ERP vendor cannot handle the basics (e.g., accounting close), why trust the vendor with higher level capabilities (AI/ML)?  Services trump software in a cloud delivery model.
  • Technologies, like business functional activities, can be siloed and experience the same limitations without having deep integrations between technologies.
  • Without global standardization for blockchains, IoT, and business process models, the most practical approach would be a single vendor solution.
  • Constantly worrying, watching, or fixing basic business transactions within your ERP cloud service will have a negative impact on the future adoption of higher-level ERP functionality.

Now, allow me to switch gears to discuss deployment considerations for autonomous ERP.

In addition to the above considerations, a strategic step to prepare for autonomous ERP is to promote and, dare I say, even demand automation across your value network (customers, suppliers, and employees). The best way to ensure that all business transactions are electronic is to do it at the source! Optical Character Recognition (OCR), like RPA, is a gap technology used to bridge the manual and electronic worlds. Additionally, this automation can reduce operational costs. However, scanning still requires manual effort on your part, is slower, and more error-prone than having a fully automated flow. Any manual efforts supporting business transactions are not flexible and are typically costly to adapt and scale. Further, removing manual steps will improve data quality, which is a core requirement for any dependable AI/ML capabilities. 

A misconception within the ERP industry concerns the role that AI/ML plays in generating revenue. AI/ML identifies possible opportunities, but it will be your people who execute to realize the wins. After all, technology is simply an enabler! Admittedly, people have the greater capacity for creative thinking and generating additional revenue, but you need to give your people the time and information for creative expression by delegating the basic business tasks to autonomous ERP. 

Summary

Dream with me a little. ERP user access is natural and consistent across all channels. There is no longer a need for sub-ledger reconciliations since journal entries are calculated and booked in real time. Accounting closure is done at the transaction level, supporting instantaneous pro forma financial reporting. Employees manage their own career paths and job scenarios. In addition, customers see your company as a trusted advisor and not a vendor. Business users spend most of their efforts on processing exceptions, making decisions based on recommendations and performing what-if scenarios with an impact analysis on shareholder value, customer experience, and regulatory compliance. Is it possible? Yes. Will it require discipline across multiple domains? Yes. Will we have to think outside of our current work experiences? Yes. Is it easy? No. Consider the following:

Until there is global standardization regarding emerging technologies (e.g., Blockchain) and business process models, your best bet is to find a single vendor that can fulfill as much autonomous ERP vision as possible, and provides repeatable, reliable cloud services to support an enterprise solution. 

In short, compared with AI/ML, the human brain is still superior and cost-effective  when it comes to creative thinking. Today, your company’s “brainpower” is limited to organizational silos and dedicated jobs. Autonomous ERP has the potential of harnessing this vast, untapped resource across all business processes via trusted, automated tactical business decisions. 

Sources

  1. An Autonomous Supply Chain: Is it Possible (Redwood)
  2. The Internet of Things How the Next Evolution of the Internet (Cisco)
  3. IoT Is Changing Everything (Cisco)
  4. 5 Key Challenges for Blockchain Adoption (Blockchain Council)
  5. Industry 4.0: the fourth industrial revolution (i-scoop)
  6. From Human Resources to People Enablement – HCM in the 21st century (InflectionHR)
  7. Time changes & the impact on ERP, accounting and business practices (diginomica)

3 thoughts on “Autonomous ERP

  1. Interesting. It is certainly something to be added, but you haven’t cleared the real barrier here – legal. Periods don’t exist because of companies, but regulations.

  2. Amazon and from the get go Uber’s internal businesses are already ‘autonomous ERP’ systems. Noticeably both are built on event store databases rather than traditional relational databases. A key reason being the separation of the read model from the write model. This allows a far more granular approach to database consistency suited to geographic scale with sub-ledgers who can come to transactional consisency without batch latency.

    Uber’s city/zip/driver/passenger/weather/traffic/event ride share model. Amazon’s autonomous fullfillment opimizations.

  3. Very interesting article. IFS has begun to lay the foundation for much of what you are describing in their IFS Cloud offering. There are some use cases in place using AI/AR/ML, but a fully enabled auto ERP across the entire platform is still in the future. The biggest challenge, as you mentioned, will be to have enough data and transactions to teach the AI and make its recommendations reliable enough to trust.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s