An old notebook
I’m not a fan of nostalgia, but sometimes those “X years ago” cheap-shots are just wonderful. Damn. Today, Google Photos reminds me, marks 1 year  since I took this photo.
At that time work was pretty insane. With only 5 software engineers — myself included — there was too much going on. We’d gotten thrown out of our previous office, working from a pretty shitty co-working until the new one was ready. Cramped, noisy, hot. Hardly the space to think things through.
Being a founder and running things meant having people interrupting you constantly. Until you’ve got enough team / structure and can delegate you make a lot of decisions, nothing surprising here. Being unavailable blocks a shitload of people. Sure, this is symptom of centralisation and poor delegation, but building a capable — and fully contextualised — team takes time. Startups rarely get to do it in their first year or so.
Hence, working on things requiring laser focus was not feasible, specially during work hours. This is why I worked from home during that weekend, getting that first draft  of the product out of my system. MGMT played non-stop. I sketched for some 30 hours that weekend. It was a lot of fun. That's when I realised working for Product might make me happier than coding.
Looking at these and reading the specs was a humbling experience. We did ship it. It was very successful, making hundreds of thousands of deliveries in the following months.
The sketches held up remarkably well. Most of it ended up verbatim on the final product: a direct result of having sweat the details out. The many iterations it took me to get there show it.
It worked a little too well. Well enough to be extended and abused into many novel uses.
Unfortunately it didn't stop us from building the wrong product. The plan was solid and we executed it well. What we didn't foresee was how narrow the information we started from was. It's not we strayed away on execution, rather we chased the wrong rabbit (pun intended).
You can always wish for broader and deeper planning. Had we surveyed the landscape better, we'd known. This makes for a nice punch line but it isn't true. So much happened between the initial sketches, at times rendering moot our initial assumptions. For instance, we designed the product around the goal of being asset light. No warehouse: all peer to peer. We know now this isn't feasible , but so was having a warehouse then. After our series B we could afford a warehouse, making a lot of the initial work matter less, but that happened nearly launch, way too late.
I started writing this hoping to arrive at a critique of poor planning. As I reflect on the path we took and the surrounding context I sense something else is to blame. We could not foresee, never mind plan, for problems that would not exist until much later.
In fact, I'm working on the missing pieces right now. With more information, actual data and deeper understanding it's easy to fill in the blanks.
Failing to arrive at an inspiring insight is frustrating, but hey, at least it's honest. Maybe, just maybe, the exercise we could have done was asking ourselves:
"What if the core assumptions are just wrong? How could they be different? How can we test them?".
Note that this is very different then validating your assumptions early (lean style). Our assumptions did hold, they were just looking at the less interesting problem. In fact, we did validate our ideas with our target users at the time. Wrong users, though. We didn't even have access to the right ones. Only after the initial, limited, version was up and running that we were able to broaden our horizons.
These questions are definitely worth asking. From now onwards I'll keep that in mind. I'll let you know how that works a year from now.
 For Portuguese speakers I can't let Leminski's haikai out:
Abrindo um antigo caderno
foi que eu descobri:
Antigamente eu era eterno.
Translating this to English would be pointless, sorry.
 I've since done quite a bit of wire-framing. I hoped practice would make me forgo pen and paper, jumping straight to software. This never happened. Sketching over paper still feels nimbler, less cumbersome and more fun than everything else. Maybe this means we lack great tools and one might appear soon. Another possibility is pen & paper are too versatile and efficient to be easily replaced. Again, time will tell.