Our Company

The new Active ERP Suites

The new generation of ERP Suites, called Active ERP Suites, incorporates Artificial Intelligence (AI) and Intelligent Robotics (IR) into the traditional ERP software.

Aboard Software’s aERP is the first such ERP Suite integrating AI and IR into a state of the art traditional ERP Suite. It is deployed and in use by multiple customers.

But why? What are the advantages of using AI and IR in ERP software? What do AI and IR do?

How do AI and IR work in the ERP arena?

What is a software robot?

An algorithm that determines what tasks need to be done and performs those tasks, like creating Work Orders or Purchase Orders is such a robot. While a software robot is not manifested physically, it is no less of a robot. It acts following certain rules to perform certain expected tasks, creating new data structures like Work Orders or Purchase Orders for example.

What is Artificial Intelligence as applied to software?

Consider a problem that an algorithm can't solve, like optimally (or even acceptably) scheduling Work Orders. The number of possibilities is far too big (like in exponential problems or combinatorial problems); they can’t all be examined to choose the best as to do so might take the life of the universe. However, an algorithm that uses heuristic criteria and certain AI methods can produce a solution that is very good, even if not optimal in the strict sense. The achievement of AI is to be able to produce a human-quality solution for problems for which an algorithm would take a forbiddingly long amount of time to find the best solution. A typical case of AI applied to software is our Operations Scheduler.

Some problems of extremely high complexity, which are impossible to solve by implementing an algorithm for them, also benefit from AI methods, because the AI solution is much simpler to implement in those cases.

Humans can produce solutions by methods that do not involve looking at all the possible solutions and choosing the best. AI tries to do the same, and in many cases it does it even better than humans.

What is an intelligent robot?

When AI is used to define complex rules for the tasks a robot performs you get an Intelligent Robot. Good examples are our robotic users, that create Work Orders and Purchase Orders and optimize merchandise utilization by applying complex heuristic rules that do not depend only on the mechanics of building Work Orders or Purchase Orders, but that also incorporate experience gathered through learning and summarized in the heuristic rules that guide the robot in order to optimize the inventory and the operations.

This is the case when a simple robotic algorithm is guided by heuristic information that provides rules that transcend the definition of the task to perform. This heuristic information could be compared to the experience that a good plant manager or shipping manager brings in when he is hired by a company. He has no specific experience with the new plant, the new products, the new warehouses and the new manufacturing processes, but he looks for efficiency and effectiveness applying variations of rules he learned to apply to other similar problems in other companies.

Refining heuristic criteria initially loaded from prior knowledge databases by adding new criteria created from the robot’s own accumulated experience in the field in order to perform a task better is more than robotics, because the robots don't follow simple predetermined rules that come directly from the definition of the task to perform, and because the robots learn how to perform the tasks better as they do their daily work.

So that is more than robotics, it is what we call Intelligent Robotics, because the robots are guided by more than the simple rules directly derived from the task to perform.

In the case of creating Work Orders or Packing Lists, for example, the key is to look for the best assignment of the available merchandise to delay acquisitions as much as possible, but still to ship the merchandise on time as expected by the customer, breaking down big production Work Orders to fit the working shifts, placing Purchase Orders at the right times, etc.

There are many ways in which those tasks can be achieved to always ship on time and delay acquisitions as much as possible.

Our Supply Chain Optimizer is a perfect example of Intelligent Robotics, it does what needs to be done, but it's guided by heuristic rules that we refined over many years of experience.

A comparative between the two methodologies

Below we detail the advantages of AI and IR by using a typical food manufacturing company as an example, and comparing what the humans had to do when using a traditional powerful ERP Suite with what the humans had to do when using a new generation Active ERP Suite.

Here’s a summary of the activity volume of a typical food manufacturing company we will use for this comparative analysis, for just one month:

Is the saved work all the advantage that AI and IR bring?

In the comparative above we have assumed that humans would perform impeccably, with no human errors introduced, and thus with no costs generated by having to identify (suffer) and correct human errors that are naturally introduced when operations are run by human users.

This is not a complete picture, far from it.

Shipping late to an important customer can affect the future sales of a company in a drastic way, orders can be lost, and much worse, customers can be lost due to late shipments.

Failure to order needed raw materials in turn can delay the execution of work orders, which in turn can lead to late shipments and idle personnel.

And so on.

There are many actions taken, or not taken, by human users that determine that the operations of the company are far from optimal. This means lost business, late shipments, late collections, impacts in the cash flow, more difficult repeat sales and a huge number of situations that CEOs had to live with when operations were carried out by humans supported by a traditional ERP Suite.

Robots, on the other hand, never forget a shipment that will be due, never forget to update the files, never forget to check for the availability of raw \materials, etc.

That is the biggest advantage of all, far bigger than the saved work, that is in itself already huge.

Traditional ERP Suites will become unacceptable once CEOs and COOs start to see, in trade shows and through their colleagues, the capabilities of the new Active ERP Suites.

An exceptional example of the effectiveness of AI applied to ERP

An experience we had with a customer is a very good example of what AI can do to produce money for the end customers.

In a food processing plant in South Florida we were given a schedule built manually for a full week of production at the plant, starting on Monday morning and ending at the end of the day on Friday.

The version of our Shop Floor Scheduler that we were running at the time (shown in the screen shot below) was able to produce a schedule that finished the week’s workload by Tuesday at 4 PM.

In short, the scheduler was able to do the same production in less than half the time, which is equivalent to more than duplicating the plant resources and personnel costs. Such a substantial expansion of the plant, in many cases is not even feasible, but could be achieved just by using the existing resources in a better way.

Alternately, you could look at the benefit as having cut to less than half the most important cost components of the Items being manufactured that week.

The advantage came from using non-trivial schedules that allow more work in parallel, utilizing resources for several different Work Orders simultaneously, while making sure that the resources will always be available when they are needed, but avoiding having “idle time” when they could be used for a different Work Order.

Think of the Work Orders as puzzle pieces that need to be placed on the scheduling timetable without overlapping, because no resource can be used for more than one Work Order at a time, but matching their jagged edges in a way that minimizes the total length of the time it takes to process all the Work Orders.

That is something that humans are not very good at, if the problem has to be solved fast and the solution reviewed frequently when conditions at the plant change or contingencies occur.

Humans will tend to produce clear cut schedule solutions and will naturally avoid going into the complexity of trying different scheduling options where Work Orders are processed simultaneously. Even if some combination of the puzzle pieces could be found that has great efficiency, the typical plant manager will not look for them, and the simple reason for that is that looking for complex but optimal solutions would take an inordinate amount of time of the human plant managers.

Exploring the many alternatives for placing the Work Orders is a combinatorial problem that humans are slow to solve. A plant manager would need to spend several hours, or even days, trying many solutions that would work, to find the best one, but that time is not compatible with running the plant in real time. Plant managers don’t have the luxury of such time availability for the scheduling task.

But as this example shows, the impact of proper scheduling goes straight to the company’s bottom line.