Product Cost Management: AI-Driven Carbon Reduction Strategies

Join us as Sasan Hashemi, Co-Founder and CEO of Tset, reveals how manufacturers are revolutionizing product cost management through AI-driven carbon reduction strategies. Discover how companies across aerospace, automotive, and manufacturing sectors are achieving 17% cost savings while simultaneously reducing their carbon footprint through advanced data modeling and automation.

Learn about the evolution from Excel-based cost tracking to sophisticated AI platforms that model entire supply chains, predict manufacturing costs, and calculate carbon emissions with unprecedented accuracy. Sasan shares real case studies including multi-million dollar savings and insights into how global events are shaping the future of sustainable manufacturing.

Resources from this episode:

• Connect with Sasan: https://www.linkedin.com/in/sasan-hashemi/
• Learn more about Tset: https://tset.com/

Listen to the Episode

Watch the Episode

Transcript

James: Hi everyone, thanks for joining us. This is James from the CTRL+LISTEN Podcast, brought to you by Octopart. Today we have a special guest, Sasan Hashemi, the Co-Founder and CEO of Tset. Thank you so much for coming on the show. It's great to have you.

Sasan: Hey James, thank you for the invitation. Thank you for having me.

James: Anytime. Do you want to start by telling us a little bit about Tset's story, what the company does, and your background in the industry?

Sasan: Sure. At Tset we focus on product cost management. We provide a self-service platform focused on very clean, centralized data management, using the newest technology in terms of automation, whether 3D algorithms or AI. We were founded roughly eight years ago. My background is mathematics. I worked on software-driven algorithms and software first for the banking and insurance industry, and then I switched into manufacturing. My Co-Founder is a cost engineer with over 20 years of experience. We looked at the sphere of product cost management and saw a very high need, whether from external market factors or simply the need for clean data to be able to run analysis and realization. We concluded that it really made sense to build a new platform, so we started that journey. We focus on enterprise customers from different industries: aerospace, defense, automotive, machine manufacturing, off-highway vehicles such as tractors and agricultural machinery. We are doing this out of Vienna, a very European company with a global footprint.

James: Fantastic. First question: is there a lot of difference between all these sectors? You work in a wide space. Do you find many similarities between these varied sectors?

Sasan: Yes and no. At the end of the day, it is about making your products more profitable and sustainable in the short and long term. Different aspects are emphasized depending on the industry. You have the purchasing aspect: when you buy a lot of manufactured parts and components, you need to have a good handle on the cost and carbon footprint. You also have the sales perspective: if you're in the supply chain as a tier one or tier two supplier, you need to make a business case, argue it, and sell it. From an R&D and engineering perspective, you want to make the right decisions that lead to a functional, profitable, and sustainable product.

So there are different angles in terms of departments, and depending on the industry, there may be more emphasis on one or another. For example, in automotive there is a very strong emphasis on purchasing and R&D. The emphasis also depends on the product life cycle. But at the end of the day, the goal remains the same. The manufacturing technologies used are often the same. There are different functional requirements, but a casting part is a casting part, or a PCBA is a PCBA. It does not matter too much whether it is in a washing machine or in a car; the requirements differ, but the base physics, manufacturing technologies, and manufacturing models are similar. There is another difference in terms of production volumes. If you make cars, you will have a different production volume than if you make planes, and that again reflects in manufacturing technologies. But again, it is a manufacturing model of components. We of course adjust things for some industries, but the core activity and the requirements stay the same.

James: That makes total sense. Thank you. I want to bring it back to a more basic level for anyone who is not as familiar with what you do. Can we start with you explaining what carbon reduction is and how companies benefit from it?

Sasan: Sure. The idea of carbon emission reduction really gained momentum through the global climate goals and different mechanisms, whether on a global, European, or U.S. level, aimed at reducing carbon emissions. This is relevant on a geopolitical level and from a sustainability point of view; it is the right thing to do. Given different types of tariffs, taxes, and penalty instruments, especially in Europe where the European Union has a strong focus on this, it becomes very relevant for the profitability of your product.

It makes a real difference whether you have 10 tons of CO₂ emissions or more. In the industry, we usually talk about CO₂ equivalents. Methane, for example, is not CO₂ directly but has a similar effect.

Measuring this and building products and supply chains that are efficient in terms of energy consumption and in terms of the type of energy consumed is very important, not only from a sustainability point of view, but directly in terms of profitability.

That is where we came from. We started with product cost management, but for us it was very important to think in terms of total cost of ownership of the product. If regulations and mechanisms translate carbon emissions into costs in euros and dollars, then this has to be part of product cost management. We normally do not internally distinguish between euros and dollars and CO₂. We talk about true cost, and CO₂ is an additional factor that will play a role in the future.

This has slowed down a bit given the current geopolitical and economic situation. But since the underlying driver of global warming will not go away soon, we see that it will come up again and in some parts may have a drastic effect on supply chains and products.

James: Right, yes. Something to think about, even though some people do not want to think about it. What does cost and carbon calculation entail? How does the process work, and what would you say are the key points?

Sasan: In the end, you can imagine it as building a manufacturing model. Take a car as an example. You say that car has a chassis, windows, an engine or a battery, and you split it up into these components. That is the bill of material. The powertrain again has subcomponents and so on. That is a typical engineering bill of material.

When you look at modeling the cost and having all the drivers, you look more at a manufacturing model. Not only what the subpart is, but also the materials, processes, and consumers that go into it. You can model that.bYou can say: this is a casting part or a sheet metal part, and first I have to cut it, then bend it, then clean it, and maybe galvanize it. These are work centers. For galvanizing, you need material; for painting, you need material. That material comes from a place. You basically model the supply chain.

Our platform allows you to do this. You can model the supply chain, and we have algorithms that, if you give them power meters about your product or a 3D model of your parts, will do that modeling for you.

Based on that, you get a very detailed model of the supply chain of your part. Now you combine it with assumptions. You say: that machine probably costs 2 million, or that machine has an energy consumption of a certain number of kilowatt hours, and that machine is in Shanghai. In Shanghai the energy mix leads to a specific amount of CO₂ emissions and it costs a certain amount of euros per kilowatt hour.

Now you have the variable cost of a machine calculated. You do that for many parts. This is where our data management is very important, and we need to be very performant and secure.

Based on that you have a very large model of your costs and carbon emissions. I always tell people to imagine it as a really big Excel sheet. That is in fact what most companies do today. We did not invent product cost management or cost modeling, but we saw that companies struggle to get on top of this with lots of Excels.

In an age where you need structured data to make the most of analytics and technologies like AI, we had pretty good timing in the last years and were able to help a lot of clients.

James: That is fantastic. To put it simply, if people are asking how carbon reduction is measured or tracked, how would you explain the final part of the process where you present to the client how they are saving?

Sasan: When you think about carbon emissions overall, you have to think in different scopes. Normally there are three scopes.

Scope 1 and 2 cover what happens in your own plants if you are, for example, a car maker, or what happens when your employees fly. That is scope 1 and 2.

In the manufacturing industry the biggest part, especially for OEMs, of your carbon footprint is either the usage phase (scope 3 downstream) or all the things that you buy to assemble your product (scope 3 upstream). We focus on scope 3 upstream.

We say: in order for you to build a plane, a washing machine, a phone, or a car, these are the components that you buy. Based on the modeling of the manufacturing of these components, this is the amount of energy and emissions going into that. Based on these assumptions, this is the carbon emitted during that production.

You can then look at that from a sustainability point of view and from a commercial point of view.

For example, in Europe there is the Carbon Border Adjustment Mechanism. When you want to import components and parts made of steel, the European Union will ask: what is the carbon emission of that, where do you import it from, and do you pay taxes on that in that country? If you do not pay taxes there, you will need to pay taxes on the import.

So it is a very real tariff that has to be calculated in your purchase process.

James: Interesting. I saw—

Sasan: I have the feeling that was too complicated.

James: No, I think that is fantastic. Thank you. It makes total sense. I saw on your website that you listed five steps companies can take to reduce costs and carbon. Do you want to run us through what those are?

Sasan: Sure. First, it always depends on where you are as a company. To make it simpler, there are two components.

You need to map a baseline. You need to analyze your parts. In many companies, this is surprisingly difficult. Accounting is on a plant or corporate level, but if you ask, "For this exact product, how much does it cost?" people struggle to answer.

So the first part is mapping the status. That means merging a lot of data: purchasing data, internal plant data, and so on, so that you have a baseline.

Then you do a high-level heat map analysis and ask: where is this coming from? Normally, according to the Pareto principle, 80% of the cost comes from 20% of the components or activities.

You then build models for those drivers and analyze how much something should cost in an ideal world, and what the gap is.

You have to distinguish along the product life cycle. There is "should cost" for an existing part: what should this particular part cost? Then there is "could cost," where you think earlier, during concept or design, about how you could change this product to have a lower cost and carbon footprint.

These are two worlds. In the "could cost" world you have much higher leverage. You can change more, but the return is longer term and you have more restrictions. In the "should cost" world you get a faster return but have fewer possibilities, because if you already source the part and work with a supplier, asking them to change the part can be very complex.

Once you have this, you go into realization and tracking. You have your ideas, analyze where the drivers are, what the potential is long term and short term, and then you work with purchasing, design engineering, or sales to actually realize it.

My Co-Founder says: "Numbers are cool and modeling is nice, but at the end of the day it's just air if it is not realized." You have to get the horsepower on the ground.

Collaboration is key, and awareness of the topic in different departments is crucial to be not only analytical, but effective.

James: Thank you. That makes sense. I like the fact that you did not just break it into five simple steps, but made the whole process make sense. You already answered part of this, but I wanted to go back over it. Why is it beneficial to begin considering CO₂ and cost optimization as early as possible in the design phase? Besides having more time to work, are there other benefits to factoring this in early in product design?

Sasan: Yes, because you have higher levers. At the end of the day it is related. First, if you work on cost, you should consider CO₂ anyway because they will be tied together going forward.

Second, in engineering you should always make conscious decisions. That means taking into account not only functionality or supply chain risk, but also the cost footprint and the CO₂ footprint.

Having that in a very early stage is a strong feedback mechanism to engineers so they can avoid costs and emissions altogether before they exist, rather than only reducing them later.

James: Again, that makes sense. Thank you. Would you be able to walk us through a couple of case studies you have worked on, just to give us some context and make it more grounded in reality?

Sasan: Sure. We have, for example, one customer who did should costing in procurement on a very programmatic level. We worked with them from the software side and the business side to go through their spend. Within the first year, they saved more than 17%, which was in the multi-million range.

Another customer in the machine milling industry worked closely with us and their engineering team on a modular platform. That is very important in terms of variant management. We added the cost function to manufacturability and modularity considerations.

We achieved very good results in driving key decisions so that the platform was not only functional and modular, but its cost was modular in the same way and did not overcomplicate the supply chain and associated costs.

Another example is a tier two supplier where, over 24 months, we increased their close rate on RFQs by 25%. They were able to offer more competitive pricing and argue their pricing better, which led to higher trust with buyers and ultimately more closed deals.

James: Thank you. This is why it is important to have these examples, because people might look at something like this and think, "What would I be saving, maybe a million dollars?" But in context it is substantial amounts of money that people could be saving by using a service like this.

Sasan: Absolutely. Sometimes it is very simple. Imagine you have a sheet metal part that you buy and do not think too much about. You buy a million pieces a year because it goes into many products you produce. That one part costs 20 euros. If you save one euro on it, you just saved the company 10 million euros.

So the impact is really high. It is also about approach. On the sales or development side, you can always argue qualitatively. But when you have models you can easily interact with and data you can access, you can directly quantify your argument and the difference.

Sometimes it is about saying, "It is more costly, but it is worth it because we know exactly how much more it costs for that function." If you do not have that, you often say, "I think it is more expensive, but I think it is worth it," without concrete numbers.

Given that in many parts of the manufacturing industry we have structural cost problems and cost-cutting, it is important to talk about concrete numbers and be data-driven in decision-making.

James: For sure. To change topics slightly, I noticed on your website you have webinars and resources. If people are going to check those out, are there any you think would be most beneficial to read or watch?

Sasan: It is a good mix. We have a white paper on CO₂, one on ROI, and several webinars with clients. Our goal was to be broad because we have a wide range of clients.

The one with AGCO is a good example. They are a client of ours and do a great job of connecting their systems and giving relevant data to decision makers at the right time. It is enjoyable to see how they think about it and about their system landscape. I would definitely recommend that one.

There is also one with Chiron, who have very different challenges. I would recommend scrolling through them. We have an abstract for each. See if you can relate to the situation or find it practical, and then watch it.

James: I want to step back from looking specifically at your company and look at some trends in your sector. What are some of the carbon reduction trends you have noticed in recent years? How have things changed in the last couple of years in this space?

Sasan: I think step one is getting a baseline. That is sometimes very difficult because some companies have hundreds of thousands of parts. Many companies started with simpler methods.

One method for analyzing the carbon footprint is called the spend-based method. You say, "I have a factor I multiply with my cost, and that will be roughly my CO₂ output." It is not wrong, because normally things that cost more have more weight and more CO₂, but it is misleading in some cases.

For example, you buy a graphics card. It is very expensive but does not necessarily have the same carbon footprint as a cast iron cube. That is something you should watch for.

Over time companies try to become more sophisticated. I think now we are at a point where they combine carbon analysis with costing, because on the costing side they already have lots of analysis. They try to derive the drivers: weight, type of material, where the material is sourced, and so on.

Some early movers were smart and tried to secure the global supply of green steel because they understood that green steel production will be limited. That is again a price delta.

This is how many companies started. It typically starts at the OEMs, either for strategic reasons or driven by legal instruments. These legal instruments are roughly defined but not really painful yet, which is a problem, but it is clear that it is only a matter of time.

The big OEMs are already there, and once you have the OEMs, they push this through the supply chain.

They then come into situations where an OEM asks its purchaser, "For all your spend, all the parts you buy, tell me the CO₂ footprint." The purchaser goes to the supplier, and the supplier says, "How should I know this? I do not have a methodology." Then they call us and say, "We need this fast."

We normally give them a fast analysis of what they need and then give them the tools and knowledge to include this in their daily operational work, so for future requests they can calculate it themselves. It is a trickle-down effect through the manufacturing pyramid from the top.

James: Interesting. Thank you. I think it is important to understand the big-picture view of the space to understand how your company operates within it.
You talked about evolving intelligence in this space. What do you see as the role of AI in carbon reduction, and how do you see this evolving in the coming years?

Sasan: AI is a big word, but I would interpret it as the newest AI advancements, which are more in the direction of text models, Large Language Models.
It will be very impactful, but maybe not in the way people expect. These models can be used for calculations, but that is not what they were built for.
They are extremely good at categorizing and understanding things. That is a big task, because every modeling of cost and CO₂ is a parametric model with inputs and outputs. Having the right set of inputs is the work you need to do: all the right assumptions.

AI will help strongly to take existing data, which is sometimes scattered and unstructured—product data, commercial data—and transform it into structured input which cost models can work with. That is something I definitely see coming.

On the output side, you have cost models for a complete product and get many numbers, not just one result. For every part you get costs and CO₂. That can be overwhelming—and that is just one product. If you have a thousand products, you suddenly have millions of numbers.

You have to understand these numbers and cut through the noise, often without deep knowledge of the models. That is an explanatory use case, similar to going to an expert and asking, "I see a gap of 30%. Can you explain where the driver is?"

Translating numbers into language and into measures—actions—is something AI can help with. The output has to be an action: what should I change. Translating data into action or insight is where AI can help. I would not recommend just dumping everything into a general model, but Large Language Models can be trained vertically or functionally for specific use cases. With enough training and use cases, this will be good.

The drawback is that these models are probabilistic. Sometimes they are not consistent. I think it is a solvable problem. You have to build in good quality assurance.

It will take time until the technology and use cases mature and trust is built. If you are an advisor, you build on trust. One really bad piece of advice can outweigh a hundred good ones.

So I do not see AI agents running around optimizing everything on their own. I see them helping experts to be more productive. I see it more as a copilot.

James: We had a guest on the show a few episodes ago and discussed how you need a human factor to balance things out, because one wrong part of an algorithm will keep producing incorrect data, learn from that data, and spiral. You need the human factor to check that it is working things out correctly.

Sasan: And of course it is about the magnitude of errors. Back to the earlier example: if you save one euro on a part, you save 10 million. On the other side, if you miss one euro, you lose 10 million.

So it is important to consider the cost of error. Especially in cost modeling and cost management, the cost of error, depending on the stage you are in, can be huge.

James: For sure. What would you say has been the effect of global events such as COVID, regional conflicts, and tariffs on the carbon reduction space?

Sasan: They definitely slowed it down. People are not saying, "This will never happen," but it is slower. Especially in Europe, the manufacturing industry is under enormous stress. Tariffs make supply chains harder and drive costs up. When you are fighting for survival, it is difficult to have sustainability as the number one priority.

This is reflected in the regulations, which are coming out more slowly. But as the optimist I am, I think the geopolitical factors and COVID will move out of focus over time.

I am a big believer in carbon emission analysis—at least that you know your emissions. I am not here to preach that you must reduce them. My personal opinion does not matter. I say you should at least know what they are. Then you can consciously decide if you care or not.

That is a long-term trend that will stay. It has a hiccup now and has slowed down, but in general, given that we just talked about AI, if you forecast how many data centers there have to be to match the amount of AI in 10 years, and if you calculate that a really big data center almost needs a whole nuclear reactor to supply it with energy, then you see that we need to be smart about carbon reduction. Otherwise the numbers do not work.

James: For sure. Thank you. I am glad you look at it with such a logical viewpoint. Last question, an easy one: if people want to follow your company or get in touch and see what your offerings are, what are the best places to do that?

Sasan: We have a great marketing team doing a lot of good work on LinkedIn, so just follow us there. We have a newsletter where we share information about upcoming events.

We have an event in Munich called "The Future of Cost Engineering" in November. We are happy to welcome you there—space is limited, so you have to be quick.

We are not really a startup anymore, but we are still a young company of roughly 60 people. We are pretty simple: you can just drop us a line and we will answer.

James: Fantastic. I can confirm this because I have done so myself to get in contact with you. Thank you so much for coming on the show. It has been fascinating talking to you. I definitely have a deeper understanding of carbon reduction and how the process works. I appreciate your time.

Sasan: Thank you, James. Thank you for inviting me.

James: Anytime. And to anyone tuning in, thank you so much for listening, and come back next time for another guest.