tf Labs 002: anything goes

tf Labs is one full day a quarter with no client work on it, and no obligation beyond turning up curious and leaving with something built. Nobody sees the theme until nine on the morning, so there is no preparing quietly for a fortnight beforehand, and by late afternoon everyone is in front of the room demonstrating something that runs rather than describing something that might.

The first edition handed everyone a demanding abstract standard and asked them to build to it. Uncanny Valley wanted the solution you would not notice you were using, which sent several pods into a long argument about what invisible even means before anyone wrote a line. It was a good way to start the initiative, because it made the point that Labs is about thinking rather than typing. It was also, by common agreement afterwards, quite hard work.

So 002 went the other way. The theme was anything goes: no constraint, no shared problem, nothing to interpret. All you had to do was prototype an entire product on the day, take a few strategic snack breaks, and have something working to show by the end of it. Where the first edition tested whether people could hold a difficult idea steady, this one tested whether they could finish.

Three minutes to make the case

The other change was at the end of the day. Presentations ran to three minutes each, which is not long to explain an idea, demonstrate it working and say why it matters, and the constraint did what short constraints usually do: the pitches that landed were the ones where somebody had decided in advance what the single most convincing thing to show was, and led with it.

Scoring leaned into the same instinct. Every idea was marked on problem solving, on creativity, and on how much fun it created in the room while it was being presented - which sounds like the soft criterion right up until you watch it separate the field. A room's delight is a reasonable proxy for whether an idea has actually landed, and it is very difficult to fake with a slide.

What people built

Gareth Hayhow took first place with a party-style maze runner. Everyone in the room joined the lobby from their own phone by scanning a QR code, then raced through a maze round after round, earning points for winning and losing them for hitting hazards. It was a genuinely good game, it was built in a day, and it had the whole room playing rather than watching, which is a difficult thing to pull off in a three-minute slot and exactly the sort of thing the fun criterion is there to reward.

Cleoné van Helsdingen and Pretty Pandey tied for second with two takes on the same underlying problem, which is that teams working next to each other do not always speak the same language. Cleoné built a gamified translator between dev speak and marketing speak, in both directions, wrapped around a Tamagotchi-style character that levels up the more the app gets used - a neat piece of design thinking, in that it makes the reward for regular cross-team communication something you can see growing. Pretty built a central feed that pulls company data and comms together from across the tools we already use, so that finding out what is going on, or onboarding onto a team, stops depending on knowing which person to ask.

Kayley Smith came third with a platform for renting things to and from your neighbours: the drill you use twice a year, the trailer, the camping gear that lives in a cupboard. It puts idle stuff to work for a bit of money, and the part that made it more than a marketplace pitch was the second-order effect she led with, which is that borrowing from the person three doors down builds a bit of community as a side effect of the transaction.

A room's delight is a reasonable proxy for whether an idea has landed, and very difficult to fake with a slide.

What the second edition taught us

The thing worth noting about 002 is that four people each carried a product from nothing to a working demo inside a single day. That is a different exercise from the first edition's pods, and it surfaced something the pod format hides: how much any one person can actually get built now, working alongside the tooling, when the day is cleared and nobody is waiting on anybody else. Both formats earn their place, and running them back to back has made the contrast between them the most useful thing we have learned so far.

It was a genuinely inspiring set of demos for the amount of time available, and the standard has moved. Edition 003 follows in September, and on this evidence it is going to be one to watch.