Tag Archives: machine learning

Could RPA Make Procurement Jobs More Human? – Best of the Blog 2019

The new “hot” technology generating hype in 2019 is Robotic Process Automation (RPA). Here’s how it can help procurement…

RPA - procurement
Photo by Matan Segev from Pexels

This article was written by Bertrand Maltaverne, and first published in February.

Procurement is, by nature, in the business of relationships. Whether it’s managing suppliers or stakeholders, the success of any procurement organisation relies heavily on building relationships between people.

Despite this, many procurement professionals do not have the time to focus on the human side of their job. Data collection, reporting, transactional activities, urgencies, etc. are all tasks that eat up their precious time. They prevent them from focusing on relationships that could generate more value and better outcomes.  

This problem isn’t new. It’s the main driver behind the constant, growing interest in procurement technologies that automate processes and increase efficiencies.

What is new, though, is the pace of innovation and the hype around some of the latest technologies.

Emerging technologies have begun to dominate discussions in the procurement space, and it has become impossible to avoid debates, articles, publications, etc. on artificial intelligence (AI) or blockchain. The new “hot” technology that has been generating a lot of hype in 2019 is Robotic Process Automation (RPA).

Before jumping on the RPA bandwagon, it is critical to look beyond the features to understand the bigger picture. In the case of the latest RPA technology that has integrated AI, it is about making procurement jobs more human by offloading even more mundane, robotic tasks to… robots!

The goal is to augment, not replace, people by combining the best qualities and capabilities of both human and machine to achieve better outcomes.

RPA: Copy/paste on steroids…

“[RPA is] a preconfigured software instance that uses business rules and predefined activity choreography to complete the autonomous execution of a combination of processes, activities, transactions, and tasks in one or more unrelated software systems to deliver a result or service with human exception management.”

Source: IEEE Guide for Terms and Concepts in Intelligent Process Automation

This technical definition of what RPA is and how it works can be summed up with a simple analogy. Imagine that you have to repeatedly copy data from one Excel file to another to produce a monthly report.

One way to eliminate these mundane, low-value, tedious tasks would be to create a macro that would do all the copy/paste for you. In addition to saving hours of your precious time over the course of the year, it would also reduce the risk of errors. This is, essentially, a simplified definition of what RPA is about.

It’s a way to automate repetitive and scripted actions that are usually performed manually by users (not just copy/paste!). It is a form of business process automation.

Typical Benefits

The typical benefits of RPA are:

  • efficiencies to free-up resources usually spent on manual tasks and re-focus them on core business (efficiency fuels effectiveness)
  • better consistency and compliance in data entries by reducing errors
  • from a system/IT perspective, RPA is a valuable workaround to break data silos. It avoids the costs (investment, change mgmt.) and risks associated with replacing an existing system or creating interfaces. RPA solutions sit on top of the existing infrastructure and simply simulate user actions to take data from system ‘A’ and put it in system ‘B’.

RPA has limitations and it is important to be aware of them and consider if the trade-offs are worth it. Some of them are:

  • RPA can do one thing and only one thing. If there are changes in the source or in the destination systems, then it will stop to work correctly
  • It requires extensive programming to ensure that the RPA solution takes all cases into account. If not, it will not work or, even worse, it will create even more issues as it is very consistent in executing rules. If something is off, the same error(s) will be consistently repeated
  • For the same reason, it is vital to ensure that processes are running well before implementing RPA

If RPA only had a Brain…

There’s no getting around it: RPA is a very dumb technology.  It does exactly what it’s told, blindly executing whatever set of rules it’s given. Such technology has been in use for years but on a limited scale.

However, with the advancement of other, smarter technologies opening up new opportunities to make RPA more useful and less “dumb,” it is experiencing a revival. AI is one of the emerging technologies revitalising RPA, and stirring up hype. These days, it’s rare to see RPA without an AI component, which has also lead to a lot of confusion between RPA and AI.

“[AI is] the combination of cognitive automation, machine learning (ML), reasoning, hypothesis generation and analysis, natural language processing and intentional algorithm mutation producing insights and analytics at or above human capability.”

Source: IEEE

By nature, RPA and AI are very different technologies:

Because most business processes require a combination of “DO” and “THINK,” newer generations of RPA solutions integrate AI components to:

  • Understand input via natural language processing, data extracting and mining, etc.
  • Learn from mistakes and exceptions
  • Develop/enrich rules based on experience

It is this new, smarter generation of “RPA+AI” solutions that has broader applications as a valuable tool for Procurement.

RPA Applications for Procurement

“It is not the type of business process that makes for a good candidate for RPA, but rather the characteristics of the process, such as the need for data extraction, enrichment and validation.”

The Hackett Group on Procurious

RPA is particularly well-suited for operational and transactional Procurement because these areas are characteriSed by countless manual activities. Here are some examples:

  • Automation & elimination of mundane tasks
    • Invoice processing: It is possible to drastically reduce efforts and cycle times to extract essential information from an invoice and perform an m-way match by using a combination of RPA and AI (Optical Character Recognition + Natural Language Processing)
    • RFx preparation: Tasks related to data collection (quantities from ERPs, specifications from PLMs or other file sharing systems, etc.) and even the drafting of RFXs can be streamlined by using RPA.
  • Data compliance and quality
    • Supplier onboarding: RPA can automatically get more supplier data or data needed to verify registrations or certifications by crawling the web or other data sources.
    • Data mappings and deduplication: RPA can be a great support in Master data Management (MDM) by normalising data (typos, formatting, etc.) and by ensuring that naming/typing conventions are respected.
  • Support to gain better insights
    • Supplier score-carding: This is an activity that requires thorough data collection. RPA can be leveraged to collect data from various sources and integrate the information into one system either for internal purposes and/or for the preparation of a negotiation or business review
    • Contract analysis: RPA can crawl file sharing systems, network disks, and even emails to collect and gather contracts in one central location. Then, it can extract key terms and store them as metadata in a contract management solution.

Conclusion

RPA, combined with other technologies, is an efficient way to connect data silos to win back valuable time. It can remove the “robot” work from the desk of procurement teams so they can focus on the human side of their job.

On top of that, procurement organisations can gain tremendous insights from implementing RPA because it can make new data digitally accessible and more visible.

However, it is important to keep in mind that RPA is only a workaround; it does not break silos like an end-to-end procurement platform would do.

How to Prepare Your Organisation for the Cognitive Revolution

Everyone procurement team is talking about AI, cognitive technology and machine learning. But for these technologies to work at their best, your business needs to be prepared… 

Image: Zapp2Photo / Shutterstock

There is a lot of talk these days about Artificial Intelligence, Cognitive Sourcing, Machine Learning, and data-driven procurement.

Almost every major procurement organisation in the world talks about how their organisation uses these tools to make decisions.

The direction of procurement is almost certainly towards data-driven decision-making.  This is a reality we all need to embrace.

I certainly subscribe to the notion that the best procurement decisions come from fact-based data-driven strategies and I firmly believe that over time, cognitive tools and technologies will become better and more effective than they are Today.

The truth is that we are not there yet.

As someone who’s industry is in the cross-hairs of cognitive technologies, I have been exposed to more than a few examples of how this technology works.

The category knowledge that these tools will draw from to generate their insights currently resides with guys and gals like me.  As such, we (the subject matter experts and category leaders) of the procurement space hold a special and specific set of keys that unlock these technologies.  It is with that focus that I would like to proceed.

In order for these technologies to work best there are certain fundamental elements that must be right in order for the tool to generate the best insights.

Good Data

Well organised and structured data is an essential foundation for cognitive technologies.

When it comes to any form of data analytics, the old adage “garbage in, garbage out” still holds .  Unfortunately, the vast majority of organisations simply have poor data.

Before you can point any cognitive tools at your data set, the data needs to be scrubbed and normalised.  This is still done manually by a team of people.  I’m sure one day this will be 100 per cent automated and perhaps technology will find a way to avoid these errors in the first place.  The fact is that whenever we receive data Today, it is highly flawed and requires weeks of work to make it usable.

Here is a good primer on data collections.

Be sure you allow sufficient time for your data to be cleansed before you deploy your cognitive tools.

Define your Benchmarks

The greatest value that AI and cognitive will bring is being able to benchmark your organisation in ways never before possible.

In a recent article I wrote on how to use bench-marking to develop cost estimates, but cost estimating is not the extent of how you can use bench-marking with AI.

Consider the value of bench-marking your organization against a competitor’s current performance.  Cognitive tools allow you to bring in publicly available information in real time.

Imagine that you are an electronics manufacturer and your closest rival releases their financial report.  Cognitive tools can seek out these reports and extract data from them to benchmark against your performance.  You can also combine cognitive tools with web crawlers that seek out competitor’s pricing information.  Without cognitive tools, this kind of information would require weeks or months of manually collecting data.  Cognitive tools allow this kind of analysis to be done instantly.

To take advantage of AI, take time to consider all the different ways you can measure your performance and see if you can come up with a few you never thought possible before.

Market Indices

All goods and services are affected by market forces. Staying on top of market indices is important for making strategy decisions.

An effective cognitive data strategy uses data from market indices.  Market Indices will enrich your own data and allow you to forecast into the future.  Adding this level of depth to your cognitive platform will reveal the actionable insights that cognitive data promises.

The Bureau of Labor Statistics is great resource for all kinds of indices.  If you are in construction, there are a number of private organizations that publish various indices to help forecast the future.  Look at the AIA, Dodge, and AGCjust to name a few.

Add market indices to your data set to enrich your analytics and strategise with forecasting.

Category Expertise

Cognitive technologies offer beautiful data outputs rich in data and content, on their own these outputs are just eye-candy.  The interpretation of that data and content must be made by skilled experienced subject matter experts.

Eventually we may get to the point where computers can read the data and a clear strategy will be automatically spit-out for anyone to act on.  Even then, how you act on the data will require some expertise.  Until such time, you must have your cognitive data interpreted by a human with category expertise.

It’s too easy for data to be misinterpreted and for an organisation to run-off in the wrong direction.  Even the most advanced Artificial Intelligence we have Today is unable to interpret the various human factors that go into strategy making and for that reason, Subject Matter Experts (SME’s) are still required.

Be sure you know that the person who will receive and interpret your data has the skills needed to execute a sound strategy.  After all the time and energy you invest in cognitive tools, you need to be sure your direction is sound.

Closing

The future of AI and cognitive is bright.  We are heading in a great new direction where information will rule.  Today there are a few trail-blazers paving the way for us all.  Those using these new technologies Today are sure to be better prepared Tomorrow as they find new and creative ways of using data to guide their business decisions.


This article was originally published on Luis Gile’s website. Check out more of his content here. 

Sign up for next week’s webinar: Clean Up Your Act! Category Management AI-Style.