Autumn offers on scent boosters and laundry perfume, new subscription and volume discounts, our perspective on responsible AI use, and the latest website and production updates.

🍂 Discounts on everything Autumn: scent boosters, laundry perfume, and storage tins
The holidays are over, and the weather immediately starts getting cooler and wetter again. Whenever the seasons change, we put our seasonal scents on offer. So now it is Autumn's turn.
🍂 Autumn
The Autumn scent is warm, earthy, sweet, and lightly fruity. It is difficult to describe, but the moment you smell it, you immediately think of this atmospheric season.
These were the Autumn offers from this mailing:
- Autumn scent boosters: 20% off with code HERFST20
- Autumn laundry perfume: 20% off with code HERFST20
- Autumn scent booster storage tin: 30% off with code HERFST30
- Clouds Laundry Sheets, 120 washes: 1+2, applied automatically
Clouds was a huge deal at 1+2. We had purchased too many G4 Clouds laundry sheets, so here was an offer you simply could not turn down: 7 cents per wash. Clouds is a fresh scent and pairs well with other fragrances.
Alongside changing our seasonal scents, we also let the website change with the season. Take a look at the homepage or the offers page.
Are you new and have not found your favourite scent yet? Try the Scent Guide. It matches your scent preferences with one of our Cosmeau fragrance profiles.

Scents we control ourselves
We have developed our fragrance profiles ourselves over the years. This means we know not only how they smell, but also increasingly what actually goes into them.
The fragrances we now develop in Europe are made according to our own requirements and restrictions. We require known and recognised endocrine-disrupting substances to be excluded. This gives us far more control than simply buying an off-the-shelf fragrance selected by a manufacturer.
But what about our laundry sheets, which are currently still produced in China?
Simple. The same fragrance, on the other side of the world.
For our fragrances, we work with several of the world's largest international fragrance houses. These companies have development and production sites on multiple continents.
Incidentally, the fact that a fragrance comes from a major fragrance house does not automatically mean it meets our requirements and EU requirements. A fragrance is developed according to the client's brief and requirements. For many products, price plays an important role in this.
That is why we did not want our manufacturer to source a similar fragrance itself.
Instead, we develop our fragrances in Europe and have the same formula transferred within the same fragrance house to the production site that supplies our manufacturer. Using the unique formula code, the same fragrance formula can then be produced there for us.
We also purchase the fragrance directly from the fragrance house. The fragrance drums belong to us, and the invoices come to us. So the manufacturer does not use a self-selected alternative, but the fragrance we commissioned.
This allows us to maintain as much control as possible over the fragrance formula and its origin, even while our laundry sheets are still produced externally.
The final step?
We will take it as soon as we produce our laundry detergents ourselves locally. Then we will not only develop and purchase the fragrance ourselves, but ultimately stand beside the mixer where it is added.
✨ New subscription and volume discounts, up to 55% off
There was always a gap between one-time orders and subscriptions. As a subscriber, for example, you could not always make the most of our monthly offers. And honestly, we thought that was rather odd: your loyalty deserves to be rewarded.
That is why we are changing this.
With all our laundry sheets, you will now always receive at least 50% off when buying a year's supply. You can mix and match scents as you like and combine laundry sheets for whites and darks.
And the longer you subscribe, the more you save. Your discount gradually increases by another 5%, and every now and then we will surprise you with a gift.
Subscribers will also get first access to new scents and our completely new laundry detergent when they launch at the end of this year.
And to top it all off, the annual deal now also includes a free storage tin worth €14,95. It feels like Christmas. 🎄
Do you not need a year's supply? You already get an excellent offer when buying three packs. And importantly, one-time orders will remain affordable with us. We are not going to raise their price to make a subscription look artificially more attractive.
Simply excellent laundry sheets with five enzymes. At an excellent price.

✨ Other products, volume discounts up to 20%
Many other products get volume discounts: 10% when you buy two and 20% when you buy four.
No discount codes, everyone now always gets a good discount, whenever they order. These discounts will also be added to the subscription options later, but they were not ready yet when we created this mailing.
Order your scent boosters or laundry perfume

✨ Responsible use of AI
Thank you for all the lovely and kind responses to our redesigned website! We also received critical questions about something that has suddenly become much more visible: our use of AI, for example in videos, Emma, and our Cosmeauverse.
And those questions are justified.
As with sustainability, we have noticed that opinions on AI vary widely. The discussion ranges from energy use and copyright to whether AI will ultimately lead to less creativity and, above all, a great deal of sameness.
Although our new website may make you think we have only just started, we have been experimenting with this internally for some time. Since April, we have also been building our own infrastructure so we can run more and more AI locally.
AI devours energy. How does that fit with Cosmeau?
AI uses energy. There is no point pretending otherwise.
That is one reason we are investing in local infrastructure. We use Mac Studios with a large amount of unified memory, allowing us to run large models locally without continuously relying on external data centres.
That does not automatically mean local is always more energy-efficient than the cloud. It depends on the model and what you do with it. The main advantage for us is that we gain much more control over which models we use, how often we run them, where our data goes, and ultimately the energy source too, such as solar panels in the future.
Open models are also becoming smaller and more efficient incredibly quickly. For more and more tasks, you simply do not need a gigantic model.
For example, we also want to use our local AI infrastructure internally as a knowledge base and a place where colleagues can ask questions. You do not need the largest AI model in the world every time you have a simple internal question.
And once solar panels are installed on our production facility, we can generate a larger share of this local energy use ourselves from a sustainable source.
Why are you doing this now?
Because the technology is developing absurdly fast.
Open models that you can run entirely on your own hardware are not only getting better, but also smaller and more efficient. For some specific tasks, they are now coming surprisingly close to much larger cloud models.
We find that interesting. Not because we necessarily want to put AI into everything, but because it is suddenly becoming possible to use and adapt technology ourselves that was only accessible to enormous technology companies a few years ago.
But isn't AI trained on artists' work?
This is probably the most difficult question, and we do not want to pretend there is a simple answer.
There is justified debate about how different AI models were trained, which datasets were used, permission from creators, and how copyright law should apply.
Models also differ enormously in this respect. Some models use clearly licensed datasets, while for others it is much less clear where all the training data comes from.
That is why we consciously try to consider which models we use, their licences, and what we use them for. At the same time, this field is still developing rapidly. So we will not claim that all the ethical and legal questions surrounding generative AI have already been resolved.
Won't everything become one big mass of generic AI content?
That risk absolutely exists.
Creating a character, photo, or video is getting easier all the time. But we think that makes something else even more important: the idea behind it.
We did not choose Emma's style because it happened to be easy to generate. We looked for a visual world that suits Cosmeau and lets us tell stories.
The same applies to our enzyme characters. Enzymes are incredibly interesting, but let us be honest: under a microscope, they are not exactly Disney characters.
So we literally gave them a face.
The characters' shapes and features are based on the function and structure we want to explain. This allows us to make some rather dry topics much more accessible: what an enzyme is, why a laundry detergent contains five different ones, and what they actually do.

So why use it at all?
Because I think this is ultimately the most interesting aspect of AI.
As the technical barrier to creating something gets lower and lower, the limitation shifts from “can I produce this?” to “can I come up with something interesting?”
And personally, I think that is wonderful.
People have been coming up with and telling stories for thousands of years. But until recently, some forms required a camera crew, animator, studio, or enormous budget.
Not everyone wants to listen to me in a video. And believe me, not everyone feels like reading my far too long emails from beginning to end either.
By using different formats, text, photography, animation, real people, and AI, we can tell the same story in different ways.
Almost everyone has watched a film and thought: I would have done this differently. Or had a story in their head that they could never create because they lacked the time, knowledge, or resources.
We are heading towards a world where more and more people can bring those ideas to life.
Perhaps our imagination, rather than our technical skill, will ultimately become the limiting factor.
I actually think that is a wonderful prospect.
After all, photography did not become less interesting when we no longer had to develop our own negatives. Instead, far more people were able to take photographs.
And where do we go from here?
We will keep experimenting.
We are not going to promise that we will always use AI well. We will probably make mistakes too. If something does not fit Cosmeau, we want to learn from it and adjust it or stop doing it.
AI will not replace real people in our communications either. You will continue to see Claire, me, and other real people. The Cosmeauverse is a new way to tell stories alongside them, not a replacement.
And ultimately, there is an even bigger reason behind this for us.
The past few years have not always been easy for us as a company. After the blow from Keuringsdienst van Waarde, followed by the investment and high fixed costs of our new production facility, we could also have decided to hit the brakes.
But instead, we are almost at the point we have been working towards for years: developing and producing laundry detergents ourselves, locally.
It would be such a shame to get this close and then stop innovating.
That is why we invest in production, R&D, automation, and AI. Not because every new technology is automatically better, but because experimentation has always been part of how we build Cosmeau.
Meanwhile, we welcome your criticism. Especially when you disagree with us. In any case, we have a page about AI with an explanation like this one, so it is clear and visible to everyone and covers the same questions. You can find it through the About Us page.
Because ultimately, it is not about whether we should use AI, but about making things we are proud of. Responsibly.
👩💻 Website update
After the first version of the new shop, we made a great many updates.
The shopping cart that opens when you add something has been improved. It is more stable, faster, and clearer.
For the large G6 packs, you can now mix and match laundry sheets more easily. There is also a returns portal where you can track the status and request a return more easily.
With so many updates, a few more bugs are inevitable. We resolve them as quickly as possible.
Next month, we are going to improve your account. It should become clearer and faster, and we will tackle the subscription system.
Do you have a suggestion, or have you run into a problem? Just let us know. It might already make it into the next update.
🏭 Production update
We received the second test machine this week. We have scheduled the first tests for next week.
The final formula is not ready yet. That is now our highest priority. Formulation and production are proceeding in parallel: we have a clear idea of the direction and are taking the final steps on the formula.
Next week, we will test a formula that is not yet final on the machine. Not to complete the product already, but to rule in or rule out a number of design parameters for the production process. This allows us to design a large part of the machine for mass production already, only locking in the final assumptions once the formula is final. That is faster than first waiting for the formula to be completely finished.
To reduce the risk even further, since this cannot go wrong, a third machine using a different process will arrive the following week. We expect both processes to work, but one may be more efficient or robust. And should one process not perform as expected after all, we will have the other one for further testing. This way, we will not lose another 8 to 12 weeks.
We are also making good progress on the formula itself. The biggest bottlenecks have been resolved. We are now working on cleaning power and how to balance it with the other ingredients.
Thank you for all your support and feedback. We are getting back to work with renewed energy after this holiday. It really is going to happen soon.

As always, thank you for choosing Cosmeau 💚
Sustainable greetings from Team Cosmeau!
Jason & Vivienne
















