As software continues to eat the world, Artificial Intelligence (AI) has entered into the manufacturing space. Yet according to a McKinsey & Co. article, few manufacturers are responding to opportunities and threats presented by the digital revolution in a comprehensive, coordinated way.
“AI will perform manufacturing, quality control, shorten design time, and reduce materials waste, improve production reuse, perform predictive maintenance, and more”
Andrew Ng, Creator of the deep-learning Google Brain project and an adjunct professor of computer science at Stanford University.
Watch this webinar recording and learn how your organization can be empowered by AI, creating an environment of Augmented Intelligence to:
ERIC ROBUSTELLI: Hello everyone, and welcome to the AI in Manufacturing – Impacts and Benefits webinar. My name is Eric Robustelli and I’m the North American Head of Sales here at Squirro, and I’ll be hosting today’s webinar. Our presenter for today is our Chief Sales Officer, Miguel Rodriguez, and he will be taking us through a presentation and letting us know how you can create actionable insights in manufacturing off your unstructured data. We do have a questions panel, so if you do have any questions during the webinar, please use that, and we will be sure to respond to them as soon as possible. Miguel, if you will, please take it away.
MIGUEL RODRIGUEZ: Eric, thank you very much. Welcome everybody to the AI in Manufacturing webinar. We will guide you through an introduction on use cases in the manufacturing space that we see in AI. We’ll also talk about the evolution of data analytics and showcase what are the cases that we solve here at Squirro.
So, to start with, just with this little picture here that some of you might know already. So, very often we see that companies have a bit of fear of innovation or they are too busy with day-to-day business, and with this picture we just want to encourage everybody once again to not forget to innovate, otherwise, we will lose the pace and competitors will take over.
The topic today is a very innovative one, it is around Artificial Intelligence. I will start with:
The Evolution of Data Analytics and AI.
We start our journey in the 1990s where the fantastic analytics tool, Excel, was introduced. Excel, as many of you know, is a solution that is used to analyze structured data. It shows you what happened, but it’s also a tool that requires quite advanced skills, and it’s more of a back mirror than a tool that shows you what is going on right now. So, to improve the usability of data that companies had in the 2000s, business intelligence solutions such as ClickView were introduced to the world and these solutions actually enabled two things; one is to see at a faster pace what is really happening at the moment, and it allows us also to analyze and slice and dice the data a lot better. Since 2017 onward, especially now in the last year, we are seeing the need of companies for actionable insights. So, people don’t want just to have data analysts in the companies that go through the data, but they want to empower the workforce and have those analytics systems to provide actionable insights and recommendations to the end users.
On the IT side, we saw a similar evolution, starting with data warehousing, where companies stored dated from multiple databases, going over to one of the biggest trends in the 2000s, which was Big Data. Big Data technology was here to help companies handle large amounts of data from various sources. And the latest trend that we have right now, and since a couple of years, is Artificial Intelligence, which is the capability of the computer, of a system, to take over human cognitive tasks and perform them in a way that delivers great value to the companies.
This is actually also the Fourth Industrial Revolution we are now in, and we see that in the B-to-C space companies have been adopting AI in a great way. Why are end users adopting AI in a straight fashion, when in the business area people seem to struggle more? This is quite simple and it’s because companies in the B-to-C space have done a fantastic job in making AI seamless. So, people are using it and they are not aware that it is Artificial Intelligence or even machine learning in some cases, because they’re simply consumers, through their phones or through their smart speakers, they use advanced technology in cars right now, or they have headphones, earphones that are able to translate from multiple different languages, or even have watches that provide recommendations. So Artificial Intelligence can be used always in the best possible way when it’s easy to use, when it’s transparent, and when it comes to where people interact with technology.
In the business world, we see that Artificial Intelligence and data itself will have a major impact. We have a statement from Forrester that states, “Companies that are truly insights-driven businesses will steal $1.2 trillion per annum from their less informed peers by 2020.” 2020 sounded far away a couple of years ago, but it’s just around the corner, it’s already next year, so it’s time for companies to embrace the capabilities they have to work with data and to generate insights within the companies.
If we take a look at AI in manufacturing, we often hear about just the negative things, about these robots, as an example the one from Boston Dynamics, that will take away all the jobs. But a lot of analysts forget to talk about the positive side of things for the human workers in the industry. So I’ve listed here four points that showcase the more positive side of the impact of AI and robotics in the manufacturing industry. So, if we look at Revenue Growth, manufacturers anticipate a 39% increase in revenue by 2020 through the application of AI and Robotics. 62% of business leaders and 77% of workers believe that AI will either help do their current job better or reduce repetitive, cumbersome tasks that everybody is annoyed to do. Also, 80% are planning to retrain or redeploy employees whom AI will displace. I think this is great news. Basically, the last point is ‘we already have been here before.’ The Industrial Revolution of the 1700s saw hand production methods replaced by machines and the factory system was born. The impact of it or the effect of it was that this transformation lonely grew production and the manufacturing workforce. So we already saw innovation in manufacturing take place many, many years ago, also with the same fear that machines would take over all the jobs of the workers in plants, but we saw the opposite. People were retrained, they got the opportunity for new jobs, and the workforce of the companies even grew and became bigger.
If we take a look at technology, obviously the field of Artificial Intelligence and machine learning is quite broad. So we see here an overview of some of the capabilities that are in the market, and more specifically some of the capabilities that we work on here at Squirro, that we have leveraged in our solution. So, at our company, we work heavily on applications in the text space, in the speech space, in the analytical space. We work on automation and recommendation. We look at the application side of things. We look also in the capabilities side, at topics around search, optimization, knowledge representation, natural language understanding, and processing. Things we have already done included within our solutions include speech to text, computer vision, and obviously from the more technical point of view, our solution can go up even to deep learning technologies.
If we look at what are the use cases that are delivered through this technology in the manufacturing space, we have use cases that have been received more actively in the last couple of years and others that are more at the beginning. So, the hottest case right now is around asset management, obviously also driven by IoT sensor data that is becoming available to companies. Other cases, specifically the ones that we also work on, like the use of Artificial Intelligence in supply chain management, in sales and marketing, and service, are trending topics right now, and very, very important, especially sales and marketing, and service, some of the cases with the highest adoption within companies, and one of the emerging cases that is becoming more and more relevant and I’m sure will become one of the hottest cases in the next couple of years is the creation of new digital services. That’s something that we have been working on already with some of our customers and we predict that this will become not only the trend now, but will become standard practice in most companies.
So, to quickly introduce Squirro to you, I’ll just use two slides. We have here an introduction to our company. We position ourselves as an Augmented Intelligence company, so we are a software company that is based in Zurich, but we have local offices, as well, in New York, London, and Singapore. We are a Gartner Cool Vendor. Counting with salesforce ventures among our investors and delivering our services through a great global network that today counts over 40 partners all over the world. What is our mission statement with the applications that we provide? We heard before around some of the application technologies that are used. Our mission statement is around enhancing, documenting, not replacing humans. It’s about providing them tangible, actionable insights in real business context, because in most of the applications we deliver, we integrate with third-party applications and then we are pragmatic about the Artificial Intelligence topic. So, when Artificial Intelligence is possible, we leverage it. So when there is enough data, when the use case is the right one, we use it. But in cases where it’s not possible because there is not enough training data or the datum is not labeled well enough, the quality is not good enough, or the use case might not be a fitting one, then we leverage other technologies, such as machine learning.
The core offering of the software that we build here at Squirro is centered around three areas. First, you can gather data from both internal and external data sources. Internal data sources can be CRM systems, document management systems, email, data from machines, data from any core system that you use. External data is information coming from, for example, a news provider, premium data provider, social media, RSS feeds, et cetera.
Once we have this data together, we analyze this data with a focus mainly on unstructured textual data. And we structure this unstructured data and then we co-mingle it, so we combine it with structured data, so the companies can leverage 100% of the information that they have. All of this is presented in a very user-friendly way in dashboards that in most of the cases are integrated into other applications and provide recommendations. This is the platform itself, which solves any problem around unstructured data analytics, also around search as one of the core capabilities and data visualization, but more specifically we also have built applications that come out of the box with prebuilt connectors, prebuilt models, and prebuilt visualization to cover specific problems.
So, what we see here is in the past, how unstructured data has been used. Unstructured data represents 95% of the data within companies, it is usually scattered, it is hard to access, and when people try to access it and to analyze it, it’s a manual process, very time consuming, and you need a lot of experts within the companies that are able to do this.
In the future, to enable the workforce, what we provide already today are applications that deliver actionable recommendations where people work. So, very easy to understand, transparency is very important, and as outlined already before, these recommendations need to be actionable and they consider all the data, not just the structured data, but also the unstructured data from all those different sides.
Here are a couple of use cases which were we provide to some of our customers all over the world.
Enhanced Sales, what does that mean? This means that with our solution we can enhance your sales team. Why do we do this? We do this because traditional client engagement only covers around 20% of the client base, so the typical relationship sales you have, we think we can generate additional opportunities by automating the client engagement and activate the 80% you just usually talk to once a year or even activate prospects that you have not met yet. This provides a tremendous impact on revenue and a strong ROI when applied. But what does it mean specifically? How do we do that and what are the recommendations that we provide?
I have here an example of how our Insights Engine works. Squirro has the capability to connect a lot of different data sources, and in many cases, we connect to data that provides us information from annual reports, from earnings call transcripts, from news, from premium data, and we have within our solution pre-trained models that are able to detect what we define as catalysts. These are events that are relevant for the sales team to start engagement with the customer. So, in this case, we have an example where a company states that they plan to invest $17 million to create more than 130 jobs. The catalyst that we have detected is that they plan to invest $12 million in machinery and equipment. So if this would be one of my customers, or one of the prospects on my target list, then I would see this on the annual report on the earnings call transcript, it would be very valuable for me. The problem is that usually, as we saw on the slide before, salespeople spend most of the time engaging with people they already know and the hottest prospects. But what if ITM is one of the companies in the 80%? Then most likely we don’t have time to read all this information that is out there and to engage with this customer at the right time. Well, if you don’t do that, potentially one of your competitors will. So, with our solution, we detect these events, these catalysts, at the earliest possible moment, and directly inform the sales team about this event so they can engage directly with the customer about the relevant topic.
This is enhanced through what we define as a Sales Cockpit. So, data of an organization is scattered, it’s siloed, especially the unstructured data, and usually also within the CRM system, you won’t find all the information that you want. So, you have to go into many different systems and search for the information many, many times, read through manually, to prepare all these meetings, and it’s just very time-consuming. So it’s clear, we can integrate into the CRM systems that are used by our customers in the manufacturing space. We can provide a full 360-degree view that contains advanced search that is powered by machine learning, that can search through all the data that you have decided to connect. We can show directly within the CRM systems documents that might be within your document management system. We can analyze call notes, we can link news and significant developments to the accounts, so you have everything within the CRM system that you need to have a proper client engagement from the right moment when you have to.
If we go on to another use case that we provide at Squirro, it’s around Service. We also saw before, that is one of the hotter topics right now in the manufacturing space, around the topic of AI. This is around what we call Data-Driven Service. Data-Driven Service is split into four different pillars. So, the low-hanging fruit usually what we define as data analytics. Companies tend to have a lot of information within their customer service systems around the incidents they have generated. It’s ticketing information, resolution texts, root cause analysis texts, texts from the service technicians when you do field service reporting, and it’s also sometimes knowledge articles that are locked in context of an incident. Unfortunately, as it happens with many other systems, the service systems are just systems of records, so we just information in there, and it doesn’t give us anything back. With Squirro, we can activate all this knowledge and in context of an incident, we can, first of all, do an analysis of all the information that is in there and provide relevant information, as an example, on why customers are complaining and raising incidents. If we have recurring topics that could be solved, or if some of the requests that we have are repetitive requests, that could be solved maybe as an example through self-service. So this is one of the low-hanging fruits. Once we have this data using Squirro, you analyze it, you will receive actionable insights that are very helpful for your customer service team.
The other one is using our solution, as we said before in the example of sales, here in the example of service, as a digital support system. So, once an incident is raised, you analyze this incident ad hoc, and we can provide the recommendation on similar incidents that we have solved in the past, resolutions that worked in the past, we can show recommendations on root causes that caused the issue, and we also can link to knowledge experts or knowledge articles, so you can leverage the full knowledge that you have within your company. The customers that have leveraged our solutions so far, we were able to reduce resolution times by around 30% which is tremendous and frees up a lot of resources, and at the end saves up to 10% in operational costs.
If we move on to the next topic, we have here Enhanced Supply Chain Management, and more specifically, Procurement. Data about supplies is usually also scattered around many, many different data sources, it is hard to access. Here again, it’s very important to understand what are the events that as an example can drive the price of a certain raw material or goods that you are purchasing. So, what we offer here is the possibility to contextualize events with price movements so that people in Procurements can make better purchasing decisions. You can also detect new influencing factors before your competitors do, as we can analyze everything that is going on in all news, in all the premium data sources, that you might not be able to do because the procurement job is very time intensive and it doesn’t leave a lot of time to spend reading through news. This is also the next point. So, we leverage all this data that is out there in real time, also to see when are your suppliers mentioned in external data sources, and when are events in context of them being mentioned, that could mean that you have a potential credit risk or supply chain risk, as there might be a strike, there might be a fire, there might be a regulation change in the country, or any other event that might put the delivery of your goods at risk.
A look at one of the larger topics that we have been covering at Squirro, we have the topic of the Internet of Things. We think that Artificial Intelligence paired with IoT will transform all the companies in manufacturing because IoT itself will generate large amounts of data. We would encourage everyone who has machines in the field to make them IoT ready, and this will generate what we define as Big Data, so very, very large amounts of information and this Big Data will become relevant when it’s used in context and when the insights are made available to the workforce, as I explained already before. And at the end, Artificial Intelligence, machine learning, data analytics per se, will then unlock the power of all this data. And we are sure that Artificial Intelligence will transform the product offering of manufacturing companies in a way where companies will start to sell new digital products and services to customers and generate a lot of new revenue streams.
Cases that we have already been covering with our customers in this space include providing data visualization services to clients to keep them in the know and reduce downtimes. Connecting IoT data to Squirro, visualizing this within Squirro and analyzing it. We have the possibility to create new preventative and predictive maintenance contracts when we have this data analyzed, contextualized, and visualized in the applications. New offerings, new consulting services can be offered through this data that were not able to be offered before. Also, the capability to contextualize service report data with IoT data to provide a full maintenance and support cockpit to your clients, will be very, very interesting, and these are all services that open up new opportunities for companies to offer better service, to offer new consulting services, and increase the revenue streams provided through these channels.
If we take a look at Data and Insights Monetization, this is a very hot topic nowadays, as well. We have here a great example of how this could work out. One of our customers, the company Buhler, a very large machine manufacturer, they have decided already 1-1/2 years ago through the leveraged capability and power of our solution, to monetize the information and insights that they have and bring a new service to the market. The service is called Safefood.ai, and it is a food safety intelligence service. What they do is they collect data from official databases, news, and social media with our solution. They apply natural language processing powered by Squirro to the data that they collect to at the end issue early warnings, offer risk analysis, and indicate also how consumers react to products, all in a very easy to use dashboard. This product that they brought to market is not only offered as a premium, but it is also a version that they sell actually now to the market, and very successfully in a very short period of time, they were able to get to 800 customers. So we encourage here, as well, companies to think about what they can do with the data that they have, what they can do with the insights that they have, and if there are possibilities to generate new products and launch them to the market. Because we think this has tremendous potential for manufacturing companies in general, to open up new revenue channels that were not there before.
So, with this, I would like to end this webinar. I hope it was very insightful for everybody. If you want to get in touch with me, here are my contact details. I’m happy to answer any question that you might have by email, also to arrange WebX or a call with you to answer any questions. So, with this, I would just like to say thank you to everybody, and you will receive a recording of this webinar to our marketing team. Thank you very much, have a great day.
ERIC ROBUSTELLI: Thank you, Miguel! That was very, very insightful, and I hope everybody can use this and take a look at the recording again, and use it as a foundation to start the visual transformation in their companies. We did receive a lot of great questions during the webinar. We will take those back and answer those via email. Other than that, thank you again to everybody for joining and have a great day.
Thursday, April 25, 2019
10:00am EDT
03:00pm GMT
04:00pm CET