The Impact Of AI Chatbots on Procurement

Gaurav Gurkhe
Feb 8, 2019 · 5 min read

Bots are already a reality for some Procurement organizations. Indeed, bots are becoming mainstream for most large international companies especially when dealing with the customer.

However, developers have found that it can be really useful for Procurement specialists. With the advancement in AI, bots that were once considered as limited and quite stupid will soon be equipped with more logic and brain-like capabilities.

There are two types of modern bots:
- Current bots that rely on a simple set of data and follow rules in order to respond to specific demands;

- Advanced bots that use Natural Language Processing, or AI. The second are the most advanced and the most versatile — as natural language processing enables them to mimic human conversational patterns.

Procurement and AI chatbots… a perfect match

Procurement specialists are surrounded with data and it becomes quite easy to spend too much time looking for an information rather than using it. We can envision that AI chatbots will enable procurement specialists to converse with their AI-powered procurement software.

It is safe to assume that through Machine Learning (ML), the bot or should I say “your new colleague” will be able learn about your preferences and your organization’s policies and procedures.

By 2020, 30% of web browsing sessions will be done without a screen. –Gartner

Machine Learning combined with NLP techniques will provide your team with the required information and exclusive recommendations based on the data.
Let’s take for example contracts management. Making sure contracts are reviewed properly is an important task for Procurement specialists. Through NLP techniques, we can extract and gather in one place all the important information related to a contract and compare with them with our organizations past contracts, current business goals and policy. The AI could
highlight key information that require human attention and flag potential issues.

The AI chatbot could answer questions such as “When will the contract with … was signed”. Through an AI chatbot, the procurement team will no longer spend too much time on contracts analysis and rather focus on added-value tasks.

Currently, our goal is to improve the learning process of such AI bots, we need to have access to a lot of data to further improve the technology and especially the recommendation systems. The more we know about your organization, Procurement as a whole and the industry, the better we can impact Procurement teams.

Let’s talk about query management. A chatbot can become the single point of contact for internal and external queries about purchase orders, invoices, and much more. No need to search through your database anymore, just ask! Several companies are already successfully using such capabilities in their Procurement portals to provide quick answers to a vast amount of queries, which leaves their teams with time to focus on more complex requests and value-adding tasks.

Limitations

Non-technical limits

AI chatbots can be used in two basic ways:
- Assisted, in which a person must a human must be involved at some point in the process;
- Unassisted, in which the bot autonomously executes an activity with no human intervention.

This difference is essential in the selection of your first AI chatbot. Based on the task you want to automate; the AI will be very different.
Based on the task, sensitive information might be involved and I am pretty sure you want to have a human at some point of the process. In this case, human intervention may be needed because the company requires an audit trail that indicates the appropriate person reviewed and signed off on the request.

Indeed, most organizations have a set of internally defined policies and procedures designed to provide a level of control over procurement activities. During my last experience in Procurement, only a C-level executive could approve a purchase over $100,000… The idea is that before using a bot, you must make sure your policies and processes are ready for it.

More effort could be required to double check bots’ work, which would eliminate the productivity improvements that were bots’ biggest selling point. And you could miss out on opportunities to effectively use bots to improve processes.

Technical limits

Voice-based conversations are the most natural ones and are also the most challenging from a technological perspective, especially in a B2B context. This is due, in part, to the international nature of business. For example, names of people or companies are not familiar words that a chatbot can quickly recognize, and to make things worse, they are often not in the same language as the one used to converse with the bot.

It is actually quite tricky to make sure all names and ways of pronouncing words are well understood by a machine.In addition to technical challenges like these, there is a more human challenge: the conversational paradox. It is the number one issue slowing down bot’s adoption rate. The paradox is that something very natural (a conversation) is done with another unusual
counterpart (a machine), which turns the experience into a very unnatural one.

How hard is it to build a chatbot?

From gathering enough data to build a solid data set to making the chatbot context-aware to
building the personality of the chatbot, there are many challenges involved in the development. I’ll focus on two main issues:

Context integration
To understand a question, you need to have an idea about the context. Well, same thing for machines. In order to add linguistic context, conversations are embedded into a vector. While integrating contextual data, location, time, date or details about users and other such data
must be integrated with the chatbot.

Coherent responses
The AI chatbot must be powered to answer consistently to inputs that are semantically similar. For instance, an intelligent chatbot must provide the same answer to queries like ‘Where are you from’ and ‘where do you reside’. The secret is to train the chatbot to produce semantically consistent answers.

Once again, the need of data is crucial. As we have seen, AI chatbots will happen in Procurement. The real challenge is how will organizations integrate them in their existing digital strategy. An AI chatbot requires some
adaptation on both sides. However, the benefits far exceed the potential implementation issues. I expect SaaS solutions to provide both the software/decision-making tool and an AI chatbot so that the procurement professionals can interact with it.

Into Advanced Procurement

Gaurav Gurkhe

Written by

Into Advanced Procurement

We are committed to the advancement of procurement linked with new technologies. Through this platform, we want everyone to exchange ideas and participate to the future of procurement. Our editors are part of the international consultancy firm called CKS.

More From Medium

More from Into Advanced Procurement

Also tagged Machine Learning

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade