Why am I building Cold Anvil?
PUFF is the answer.
If you’ve not seen my other posts. I’m building Cold Anvil, it’s a development studio shipping 3 products over the next 3 months.
Why am I doing this? Am I insane? Both questions I’ve asked myself.
To try and prove my sanity, I’ll break down something I’ve called PUFF, starting with the P.
Portfolio - The age of show, not tell.
It’s never been cheaper to build. The cost of getting something real out the door has collapsed, and as a result showing off that work is easier than ever.
That can be tricky when my best work isn’t always something that can be shown publicly. I lead an IAM team at Spotify, building products that touch 750 million users. Identity, authentication, authorization, security; Foundations that set the tone for the entire product.
Serious business that often has to live behind the curtain.
Cold Anvil, gives me freedom. Freedom to share, try, learn and fail. With products that I typically wouldn’t touch in my day job, that are out of the scope of my team or out of the scope of my role.
I get to build a product portfolio, in public, end-to-end. I get to practice and showcase how I think about products in a much more general sense.
Valuable for Spotify, but also valuable for any future employer.
Understanding - Learning and failing, fast and frequently
I started my career as an apprentice, I have close to zero formal qualifications. I’m a firm believer in learning through doing.
Whilst I use AI a lot at work, much of my job (rightfully so) does not require novel AI usage. Naturally, this slims down how much hands on AI engineering I’m a part of day to day. I expect that one day, this will change, for all PMs. However, I don’t want to sit and wait for that day.
Building harnesses, training models, making tiny rubric changes that fuck up my eval architecture. This is all stuff I get to experience by doing, at a scale far larger than I would in my day job alone.
Frontier
This one is dead simple. I’m a nerd, I love technology and I adore anything that feels like science fiction. In 2008 Iron Man’s AI assistant felt like fiction. In 2026 I have my own assistant (Annie) controlling parts of my life.
The world is still very much figuring out the role that AI is going to play. We’re in the wild west. We’re at the frontier.
Each week something new emerges; Models, architectures, techniques, applications, risks.
I don’t want to just watch it play out, I want to be part of it. My day job, but especially Cold Anvil, gives me that experience.
Any opportunity that takes me further into the frontier, I’ll jump on eagerly.
Fun
As a kid, I used to make paper prototypes of software, gadgets and hardware that I would dream up.
In my teenage years, I used to write scripts for a game called Runescape, so that I could abuse automate in game tasks like ‘mining’ and ‘woodcutting’.
Why? Because inventing things to solve problems is fun.
AI is a magic wand that can turn my paper prototypes into real products. For anyone who can relate to the weird kid I described above, I know you understand how magical it feels to use these tools.
What if it fails? Spoiler: It can’t.
I’m investing a huge amount of personal time on all 4 of the products that will come out of Cold Anvil Studios.
It can’t fail, because it’s already rewarding me.
I’m building a new portfolio of experience. I’m learning every day. I’m operating on the frontier of technology. And for the first time in a while, I’m having fun.
None of this means I don’t want to succeed as a business. Of course I do, I’m not insane. My first project is filling a gap that Lovable and any other AI tool aren’t filling today.
I’ll share more about that, including a preview, in my next post.