Although obsessed with the idea of human flight, Leonardo da Vinci — whose understanding of aerodynamics was centuries ahead of its time — got only as far as the technology of his day could take him.
Right Place. Wrong Time.
His ornithopter was so close. Da Vinci had sussed out the principle of lift, but since human muscle alone can’t produce the specific power required for takeoff and the internal combustion engine was 400 years away, his work never went beyond notes and sketches. Like an image stuck at 10% download, Da Vinci could see the big, blurry blocks of fluid dynamics, airfoil shape, and leading edges; but the pixels of power-to-weight ratio, lightweight materials, and 3-axis control were still in transit.
It’s a historical heartbreak in hindsight when someone stands on the verge of changing the world, but the technology to achieve it is just out of reach.
Right Place. Right Time.
However, occasionally the stars align and great ideas appear just as the technology to execute them comes online. The two react like white phosphorus and air, igniting an advantage that continues to create its own heat indefinitely. For enterprise resource planning (ERP) software, modular AI architecture is the technology and Dynamic Work Design (DWD) is the idea.
In this article we’re going to use the example of Dynamic Work Design to illustrate how management principles that have been refined over years can be executed within the walls of the ERP, at the speed of business, and with less hands-on human intervention than anyone thought possible.
Aggro to Ally
In There’s Got to Be a Better Way: How to Deliver Results and Get Rid of the Stuff That Gets in the Way of Real Work, Nelson Repenning and Donald Kieffer, the architects of DWD, argue that software is ossified and unable to match the pace of business. They give this example:
“At one BP refinery, one-third of purchase orders were delayed for days because they required rework. The procurement software was no match for the dozens of steps necessary to request, order, and pay for parts, and it could take days to get details from refinery operators, who were rarely in front of a computer. The answer in a case like BP’s isn’t likely to devote resources to redesigning already complex computer systems.”
They believe that enterprise software is inherently rigid, no match for modern work that is dynamic and requires constant adjustment, and often frozen in a state that reflects how work was done at some point in the past.
On that, we agree with them 100%. However, on their previous point we disagree. We believe that software can be developed to evolve at the same pace the business evolves — using a modular AI architecture. We argue that IT can be a strategic ally of Dynamic Work Design by using a very specific type of AI augmentation to allow enterprise software to adapt to business needs on a daily basis.
Tech Catches Up
In a previous article on the topic of corner cases, we discussed the missing capabilities of the ERP and the operational triggers that force workarounds, leading to workflow process gaps (WPGs). We also explained how AI augmentation can respond to non-standard processes with modular AI processing units operating as a networked decision engine; identifying bottlenecks and allowing processes to be optimized quickly through existing business rules.
One of the knotty problems of modern technology is synchronizing human effort, which is mobile, with the fixed nature of enterprise software. The solution that DWD provides bypasses the software and allows the business to rapidly adapt to changing circumstances. They propose five principles for accomplishing this in a way that restores the visual, physical, and material nature of getting things done:
- Solve the right problem – focus on the business goal
- Visualize the work – use shared real-time displays
- Connect the human chain – design communication escalation paths
- Structure for discovery – turn errors into learning
- Regulate for flow – balance work against capacity
When software is taken out of the equation, workarounds are the means by which you swat corner cases away like flies. Spreadsheets and post-it notes become tools instead of contraband. However, if the human and financial metrics are all expected to live in the ERP, what happens when a transaction is inevitably not recorded? The answer, of course, is a gap between what is known and what is actual.
We admit this has been largely unavoidable — until now — when the technology that can give human adaptability to enterprise software has come online by means of artificial intelligence.
Currently, the biggest obstacle to implementing agentic AI in the enterprise is its lack of persistent memory. An AI system that could learn and remember would be able to apply the structured problem solving of DWD, executing some in the background and bringing humans into the loop when approvals are required. Once the problem has been resolved, and the process has been updated, it can be executed by AI moving forward.
Modular AI architecture executing DWD principles would be the combat multiplier those principles need to achieve their full potential and operability within ERP software. Before we talk about where the framework can go, let’s talk about where it came from — system dynamics.
The Dotted Line From System Dynamics to AI-ERP
The intellectual history of DWD goes back to the 1960s and the work of Jay Forrester. Forrester was the father of system dynamics, a field that applies engineering principles to human systems. He used the language of stocks and flows; where a stock is something that accumulates — such as inventory, people, and tasks in progress – and a flow is the rate at which stocks increase or decrease.
To illustrate how this works, let’s use the example of an operations division that rents equipment to construction jobsites. In this scenario, “flow” is the logistics and dispatch of assets (the physical movement of materials between the warehouse and the jobsites). When an asset is offline and in use at a jobsite it becomes a “stock”. In a very simplified version of the process, a piece of equipment must be mobilized and transited to a jobsite (flow) where it remains for as long as it’s needed (stock). Afterwards, it is demobilized, inspected, and checked back into the system. There are financial transactions associated with each of these activities that must be accurately accounted for if the company is to remain profitable.
In this operating model, tasks in progress aren’t complete until the equipment is removed from one project and available for the next one. As long as an asset is in use at a jobsite it’s a stock. The activities that enable it to move from one jobsite to another are flows.
Stocks and Flows Meet ERP Workarounds
Let’s look at a real world example. We have 100 pieces of equipment that are no longer needed at their current location and have other jobsites waiting for them. In the perfect world of the ERP, all 100 pieces of that equipment flow back to the warehouse, undergo inspection, and flow to their new locations. But, because this is the real world; 10 units are needed by the end of the day at one jobsite and if those units aren’t there before the start of the next shift, money will be lost. Lots of it.
The obvious solution is to have the equipment inspected forthwith at the first jobsite, and immediately mobilized to the new location. Because employees live in the real world and realize how ridiculous it would be to lose a boatload of money just to make the ERP happy, they inspect the equipment that day and get it where it needs to be on time; while making arrangements to send the 90 remaining units back to the warehouse. Hello workaround.
Once that happens, as far as the software knows, 90 pieces of equipment are on their way to the warehouse and 10 units are still at jobsite #1. Team members resourceful enough to get a million dollars worth of equipment moved in under 24 hours will record the transaction somewhere, but who knows how long it will be before the software is updated — if ever. At this point, the ERP has ceased to be a source of truth and has instead become a source of confusion.
ERP vs. IRL
When corner cases are resolved outside of the software, they haven’t gone anywhere. They still exist and they still aren’t being resolved at scale. The local fix doesn’t contribute to the knowledge and agility of the entire enterprise and the data from that fix is invisible to decision-makers.
While this situation presents a significant challenge, it’s not insurmountable. The core principles of DWD can target the root causes of process gaps efficiently and productively, if they can be executed within the walls of the ERP.
What’s needed is a way to solve problems and have the software adapt to it on the fly so that the updated process is integrated into the entire ERP and can be executed by it from then on. When this happens, the process gap will be eliminated and will not be able to generate future corner cases.
Modular AI Augmentation Enters the Chat
Modular AI augmentation layers high-impact AI modules – such as predictive analytics and intelligent automation – directly into existing processes. This allows organizations to keep their core system of record while adding new ways of working that are faster and more agile. The agents both provide insight and execute tasks. They can orchestrate processes across multiple business units, turning the static ERP into a dynamic tool.
| DWD | WPG | AI-ERP |
|---|---|---|
| Solve the right problem | WPGs arise because the software forces the business goal to bend to its will instead of vice versa | Combines deterministic and AI programming to allow the system to be as flexible as the business |
| Make the invisible visible | WPGs are uncharted terrain living in shadow ERPs | Flags the WPG and brings it out of hiding |
| Connect the human chain | The problem solving for WPGs is executed in isolation | Information and approvals are sought from humans as needed |
| Structure for discovery | High touch learning doesn’t scale when WPGs are multiplying rapidly | Is autodidactic – integrates new information and improves over time |
| Regulate for flow | WPGs take work outside the ERP so work and capacity can’t be balanced | All work stays in the ERP and can be resourced accurately |
DWD Meets AI-ERP
By now, you may be wondering what happens to Dynamic Work Design if the majority of the in-person interactions and concrete visual management systems are rendered unnecessary. We can answer that question with an example from science. In biology, if a species becomes simpler to survive better, it’s called progressive evolution because the net effect is positive. Think pandas. They’re able to monopolize a limitless food source, yet their essential bear nature is the same even though they are no longer carnivores.
If a system can interpret its own state and self-correct; the mechanisms change, but the underlying logic of its principles remains the same. For example, turning errors into learning involves reflecting on mistakes to improve a process. If the system does that without a human, we can describe it as evolutionary architecture. It becomes antifragile because every time the system closes a process gap and learns from it, it grows stronger.
Removing the human where human intervention is unnecessary doesn’t change the base principles of DWD, it reimagines their execution.
The Image below shows how DWD might be executed as conceived for the equipment transfer scenario we described above, and compares it with how they could be executed within an AI-augmented ERP. The corner case goes from needing multiple people and multiple interactions to being resolved by one person with the proper authorization; while maintaining the integrity of the DWD canon.
Transitioning from Dynamic Work Design on the floor to Dynamic Work Design in the cloud is more than just a technical upgrade; it’s a tectonic shift in the ability to get things done. By removing complex workflows from the constraints of deterministic code, a business can replace rigidity with adaptability and get rid of its cache of corner cases. DWD no longer needs to tiptoe around inflexible software demands. It can have the visibility, agility, and speed it needs for successful execution.
TL;DR
The true power of modular AI augmentation lies in its autonomous problem solving capabilities. When the system is able to perform multiple activities in the background before bringing a human into the loop, team members are free to concentrate on mission critical priorities. Ultimately, AI augmentation doesn’t replace humans, it gives them the bandwidth they need for high productivity.
Atigro CEO Ken Fischer sums it up this way:
“AI-based ERP augmentation isn’t just a tactical fix – it’s a strategic platform for continuous innovation. As new AI capabilities emerge, they can be plugged in without disruption, keeping your systems – and your competitive edge sharp.
The result is a fundamental shift in what an ERP can be: not just a system of record, but a system of intelligence. One that actively helps shape decisions, drives efficiency and does its part to future-proof the enterprise. That’s not a migration project. That’s a transformation.”
Enterprise software in the 20th century was limited in similar ways, although not to the same extent, as the ornithopter was in the 15th century; making operations management of large-scale enterprises almost as difficult in its time as flight was then. At this juncture, the timeline from ideas to capability has been compressed from 40 decades to four. In the 1990s, the map the software provided bore only a partial resemblance to the territory, making physical scaffolding necessary. Today, with AI-driven ERP augmentation, accurate information can move from the point of reality to the point of decision in seconds.
If we choose, we can evolve from executing slowly and with distortion to pivoting live with absolute clarity.
About Atigro
Atigro is a proven ERP transformation firm that pairs its modular augmentation capabilities with AI-native frameworks. Atigro’s experience and expertise generate the rapid development and provisioning of new ERP functionality that meets dynamically changing business processes. You can learn more about implementing strategic AI capabilities to substantially improve business operations throughout your company by streamlining, automating, and optimizing workflows.