I’m always amazed by the power of smell. Some smells are radically wired into our brains, connected to memories, events, and the people we love.

Easter mornings to me are the smell of “pastiera”: a traditional Italian sweet from the city of Naples, where my mum was born. That smell has the power of bringing me back to my family home’s bed during Easter mornings, where I’d be woken up by the smell of my favourite cake being baked by my mum since the early morning.

My mum’s serious pastiera-making skills

Neapolitans are really serious about their pastiera. It’s a cake that can be prepared and eaten only during Easter, no matter how many times your son asks for it for his birthday (trust me, I’ve tried). I always thought that Neapolitans were so strict in order to keep the tradition intact, but now i guess it’s also because it’s an incredibly long and tiring process that requires tons of ingredients (butter, sugar, flour, eggs, cinnamon, ricotta, wheat, orange flowers extract, caramelised orange and lemons).

This year, I decided to embark on an adventure. Living in Copenhagen and kept away from my home country and my family by a global pandemic, I decided that it was about time for me to take over my family tradition and, for the first time in my life, try to bake the cake of my heart.

Armed with my nonna’s old recipe and 24/7 Skype support from my mum, I started baking. I decided to bake three cakes. Started at 9am on Saturday, finished 11 hours later.

This was the result:

The result of my 11hr effort

OK, I cherry picked my picture. Because out of the three pastieras I baked, not all of them turned out as planned. This other one was unlucky: the small lines of dough that form the pastiera pattern drowned into the filling (can’t blame them, that filling is delicious).

Another one of my pastieras. Let’s say not perfect…

Perfectionism is one of my biggest flaws (or virtues, depending on the situation, I guess?). No matter how well I perform in anything, my focus is always on what could have been done better. So even if the cake turned out so well that my friends finished 2 out of three cakes in 15 minutes, I immediately thought:

How can I make it better next time?

I already had some ideas: use a different flour, more eggs in the dough, adjust baking time to pastiera size, don’t let the dough decoration drown.

Even if I thought that I had successfully debugged my pastiera flaws, a terrifying thought occurred to me. If I want to keep the tradition intact I can’t try perfecting it again until next Easter. All these learnings need to wait 365 days before being applied. Will I remember what I did wrong? Will the next ones be better? Will I make the same mistakes? Or maybe I’ll run into new ones? All these questions will remain unanswered for a whole year.

For a living, I help organisations understand and use Artificial Intelligence. One of the things I like the most about my job is that often I get to help people that are doing something for the first time. There’s some excitement in seeing people testing themselves, exploring new opportunities and new careers after maybe being too comfortable for too long.

Like me with my first pastiera, people want their first AI project to be perfect. They want their algorithms to work perfectly on the first try, end-users to crave the new product, and every task finished on-time and on-budget as planned well in advance.

Well. That’s unlikely. It’s hard to predict how your algorithms will perform, what users will think about your service, or if your first dough will fall into your cake’s filling (and by the way, this is true for innovative projects in general, I believe it’s not AI-specific even if AI projects do have large uncertainty).

When you’re starting a new adventure, if your first KPI is “ROI” or some Machine Learning metric, you’re setting yourself up for failure. That’s why when I approach a new AI project, my first KPI is not model accuracy or some other fancy Machine Learning metric. The first KPI I want to optimize for is “Time to First Learning”: how long does it take to develop something that can give me insights on whether I’m on the right track, or I need to adjust my strategy.

If you really believe in what you’re doing and are in for the long run, there’s no failure in experimentation: you win, or you learn. That’s why I’m not bothered by my cake’s imperfections, I’m bothered by having to wait a full year to reiterate. You should have the same mindset with your innovative projects as well: don’t seek perfection, seek fast learning.

Want to know if your cake is good? Taste it. Want to know if your project works? Test it. Hopefully you won’t have any weird tradition that prevents you from trying again until next year.