The returnable drinks container is one of the earliest examples of the circular economy in action. In the 1970s, two Norwegian brothers set out to automate the process of returning containers for reimbursement, creating a reverse distribution machine that quickly spread to stores across Europe and later around the world. The company they founded, Tomrais now a global leader in the collection, sorting and recycling of technology-based products and packaging.

Today, these reverse vending machines are digitally connected and can ingest a wide range of cans, glass bottles and plastic containers, automatically identifying, cleaning and sorting them for later collection and recycling – or rejecting those which are not acceptable. Their installation and operation requires an efficient field service operation, because if the machines break down, customers lose confidence. Daniel Basile, VP Field Service for Tomra North America and a 27-year company veteran, explains:

We have two tiers of customers. We have the retailers themselves and then a consumer comes in to return their containers. I did some customer surveys. If they go there to the location and that car isn’t ready to take their container, there’s a 30% chance they won’t go back there and do it again.

So from a two-tier customer value, from the consumer, they want to go there, walk into that room, be able to take all the containers right through the machine, walk in, continue shopping. From the retailer’s perspective, when that customer can go in there and do that, that customer doesn’t complain to the retailer – the best service is the service unseen, right? So if we provide a service that is transparent to the customer, and they don’t have to hear any complaints, that’s a lot of value that comes to that customer.

There’s a lot going on behind the scenes once a consumer has dropped their used containers into their car and received a cash refund. The cars are emptied regularly and the contents are taken to a dealer for recycling. Tomra has to track and manage the various financial transactions in progress. Jay Sethuraj, VP Technology at the company, explains:

It’s an IoT-enabled platform, so all transactions, every container, are routed to a data center. We pay retailers. Then we grant the distributor: “You wish so much to the merchant.” A lot of financial transactions take place. Therefore, turning digital results (into a transaction) accurately and securely is key to our business.

Upgrade to IFS Cloud

Sensor data is collected by servers running on the Azure cloud platform and connected to Tomra’s ERP system, recently upgraded to IFS Cloud, which also runs on Azure. The company had previously been running on an on-premise instance of IFS that first went live in 2014. In 2022, the decision was made to move to IFS Cloud, which was introduced only recently. Sethuraj explains the rationale:

Even though it was new at the time, I could clearly see that the open AI architecture and the ambition that IFS put into it was the right mindset and… that it could scale to what we saw two years later.

One of the most striking differences was the “evergreen” upgrade cycle, where new features are released several times a year without the need for the disruptive installation of an entirely new version. The latest feature build was given a 24-hour window and ultimately took only five hours, including all testing and validation, with no downtime. Sethuraj says:

The best part is that we don’t have to do big upgrades anymore. When we launched in September 2023, we were on version 23R1, and since then we’ve had several package releases, but mainly we’re (updated) within 24 hours at most and without any disruption to our users. Of course they have to do some testing and stuff like that, but nothing compared to how you upgrade from on-prem eight to nine or nine to 10.

The new system helped streamline processes for the finance team with more accurate transaction coding and simpler consolidation. Christine Fonseca, Corporate Controller, says:

It really helped my team shorten the closing process and consolidation period. We have a Canadian company and a US company in different currencies, so the ability to consolidate quickly has really helped us in the cloud environment.

And then we have real-time visibility into our transactions. (We have) the ability to be able to create P&Ls and balance sheets and slice and dice them in a number of different ways, like business unit, looking at different revenues, and (we’re) able to show management where we are at any given time in -a quick way so they can make decisions and trust the data we provide. So from a financial accounting perspective, we’ve seen a lot of benefits from IFS now certainly being in the cloud.

In field service, better visibility into various aspects of the operation helps improve metrics that are already outstanding compared to the industry as a whole. Basile comments:

Equipment is equipment, but the service we can provide behind that equipment can be a real differentiator between our customers choosing our equipment and our competitors’ equipment. Let’s get into a system where we have better metrics to understand how we’re doing today, what we’re doing well, what we need to do better at — and then have tools that give us the ability to focus those transitions on how it goes from good to excellent and keep improving it.

The impact of AI

These field service figures include an average repair time of just over 20 hours and a 96% success rate in getting machines back into service after a breakdown, either same day or next day, along with a repair for the first time in the nineties. Basile gives an example of how keeping numbers helps the company manage inventory so that repairable parts are back in service when they are needed:

We are able to understand if we need 70 printers for next week and we have 40 new ones in stock but we need to repair 30, the workbench helps us with the repair orders, with our shop technicians. , so we can make sure we have the necessary parts needed.

We have to get the technician there at the right time, in the right place, with the right parts to be able to do that job. Using (everything) with that, with our optimization, with routing, that’s how we’ve been able to hit 97-98% first fix rate.

Tomra is now working with IFS to find new ways to improve performance using AI. The vendor is working on an AI-assisted feature for its asset performance management app to enable predictive maintenance. A model is trained on two years of Tomra sensor data to be able to predict component failure from IoT data so that preventive repairs can be made. Another AI application ingests a mix of unstructured and structured data to create a virtual assistant to help service technicians. Sethuraj explains:

We have all the service manuals, user manuals, software documentation, it’s called unstructured data and it’s used to train virtual assistants for service technicians, with the intention of further training that model with structured data that comes directly from task surveys work and other (records).

In the future, Basile sees the potential to identify and share best practices within the field service team. He says:

I’m looking for ways to not only be able to look at our KPI metrics, but also get into the quality of service that was delivered. Using a lot of work order data coming into IFS (we could look) at “Hey, our best technician does things this way. How can we get this tribal knowledge of our more experienced technicians to be shared with our newer technicians?

On the finance side, Fonseca is looking forward to the impact AI can have on further automating processes and monitoring KPIs to help improve areas such as collections and inventory management. She adds:

We’re excited to see where this AI can take us. It’s an accountant’s dream to minimize our closing at the end of the month. Accountants always close the books every month, every year. The dream would be to have this done automatically in the system, maybe with just a day of checking the actions, making sure things look good.

I think we can really benefit from financial forecasting as well. Right now, we’re very tight on budget. Everything is based on Excel, several different business units and it gets very detailed and cumbersome. So we look at IFS to predict what our budget would be based on the historical data we have there. And then we might say, “Oh, we know this is going to change 10% up, down, or if there’s a new business,” but the hard work is done for us. I think it would be very useful.

For Sethuraj, there’s also a bigger picture of using insights from Tomra data to inform the industry as a whole:

We need to look at the full loop. That’s where we want to invest through AI. Our customers are generally retailers, chains as well as distributors and manufacturers. They all have their own sustainability goals, like for the Pepsi and Coca Cola industry, they wanted to bring all their containers back.

We could give them information about their sustainability mission – KPIs and things like that. We have over 15 years of data in our system, plus 10 years of service data using IFS. How do we bring everyone together?