Last Friday we hosted an AI Hackathon at 60 Decibels, and it was a roaring success. In case you want to borrow the idea, here’s a look into how we did it.

First, the why: getting our team to go from asking questions of AI tools to getting these tools to solve real problems and do real work for them. My impression is that 80%+ of our team (including me) were in the first camp, and our goal was to jumpstart getting 80%+ of our team in the second camp faster.

We knew from experience that what matters is doing, not talking or reading or watching.

We laid out dedicated time for this: a half day starting AM, with three different groups to account for time zones (India / Kenya + Europe / Americas). Each region had a leader, who was someone who had more experience with AI tools and could serve as an internal expert / troubleshooter / problem-solver, as well as a second facilitator. Each team worked together, on Zoom.

The logistics were important. Here were some elements we put in place:

• Dedicated time per regional team
• Crowdsourcing ideas of things to work on in advance (Google sheet) — anyone can add
• Advance sign-up to confirm participation (Google sheet), and confirmation if people wanted to work alone or in a group
• Zoom call for each of the regional teams, with a everyone in the same room for the introduction and ‘idea pitch’, then breakout rooms for each team. The ‘experts’ were always accessible to answer questions during the session in the main Zoom room.
• We set up a dedicated private Slack channel for the planning group (~8 people), about a week before the Hackaton, and repurposed another public AI channel for all other comms and day-of communication
• We got IT issues / logins sorted in advance: made sure we had enough pro accounts across the tools we wanted to use (primarily Claude); confirmed logins available to these tools; confirmed that a person in each region had the admin access to buy more tokens if we ran out / access to extra accounts if something went sideways.
• Worked with our IT support to pre-load whatever apps we needed onto people’s laptops before the Hackaton (again, primarily Claude), because many people don’t have admin rights to download new software
• (aside: a few folks did get ground to a halt on the Claude Cowork installation on PCs the day of…I assume Claude will work this out shortly)
• Reminded our teams in advance of our safe use policy for all tools

We created and shared a document with background reading—definitely the sort of thing AI can create for you. It covered everything (agenda, safe use policy, ‘What is AI really?’, Choosing the Right AI Tool(s), How to Talk to AI, Agentic AI for Beginners, Suggested Reading). The suggested pre-reading / watching was:

• The Future of AI at Work: Introducing Cowork – (Video) Webinar recording
• Navigating the Claude Desktop App – (Article) A helpful comparison on when to use each Claude tool
• Identifying and Scaling AI Use Cases – (Article) A short guide to help narrow down on use cases for AI across different functions
• Claude Tutorials – (Article) Practical examples on the application of AI in real work scenarios.
• Something Big is Happening – (Article) How AI is truly revolutionizing the way we work (and live!)
• AI Agents, Clearly Explained – (Video) 10m explanatory video
• Master the GPT Prompt Formula – (Video) How to write an effective prompt

In terms of what happened, the team did incredible things.

• They generated 58 ideas of meaningful business problems to work on, and worked on 20+ of these across three time zones.
• These were big and small—many at the core of some of the most difficult problems we have as a business, some much smaller, nagging issues that bug just one person.

Best of all was seeing the energy and the enthusiasm of the team, people jumping in to help each other, people feeling empowered to tackle issues themselves or in teams. We’re definitely going to do it again!

At this point, you’ll have seen or heard about this bumper sticker.

This flavor of virtue-signaling is particularly ham-fisted, but it’s also an example of something that happens all the time.

What interests me about this is the ways in which our likelihood to act is impacted by our internal narrative.

“I bought this before Elon went crazy” communicates, presumably, some sort of solidarity along the lines of, “Just so you know, I’m not in favor of the authoritarian tendencies, the dehumanizing of people, the willingness to pull billions of dollars of life-saving spending on a whim. I’m a good person just like you.”

That last sentence is likely true, but the much more accurate sentence is, “While I’m sickened by what’s going on, that unease is less important to me than the inconvenience of selling my $80,000 car to buy another $80,000 car.”

And that’s really the rub: we plaster the sticker on the car, and it serves as a release valve, in the form of saving us from an uglier story about ourselves than the one we’d like to believe.

We repeat this pattern in hundreds of smaller ways. It shows up every time we talk about something being broken without putting ourselves on the hook to fix the broken thing.

“Coming into the office more doesn’t make sense because we just take our calls from the office instead of from home. No one takes advantage of the in-person time.”

“I would love to advance in this company, but I can’t because I never get any feedback.”

“People on my team don’t take responsibility for their actions. There’s no follow-through.”

“My kids are addicted to their phones; it’s impossible to get their attention.”

“The culture around here doesn’t make people feel valued for their unique contributions.”

“I just can’t get this build right because the user requirements aren’t clear.”

“Morale just isn’t what it used to be. People aren’t feeling a sense of connection.”

“We’re just not taking advantage of all that AI has to offer. Everyone is being so timid.”

When we articulate a concern about the current state of affairs, we might be doing one of a few overlapping things:

1. Group creation: by signaling that I see the world the way you do, I create solidarity with you
2. Narrative to self: by naming something I dislike, I maintain my self-image as someone who doesn’t support that negative thing
3. Venting: getting something off our chest so we can get ourselves unstuck
4. Persuasion: attempting to convince others that something is not the way it should be (ideally to enlist them to make change)
5. Exploration: engaging in dialogue to figure out more clearly what’s going on that needs to be fixed
6. Enlisting authority: we inform someone with greater ability than ours about a problem that we’d like them to address

And while it’s true that all of these actions can be done with more or less intention to act, typically the first three decrease pressure to act (because they are focused on internal / shared narrative) and the second three increase pressure to act.

For example, our favorite bumper sticker:

• Group creation: identifies us (to ourselves, others) as anti-Musk
• Narrative to self: “I’m a Tesla person with an anti-Musk bumper sticker. That’s a little less uncomfortable.”
• Pressure to act: goes down, because I feel like less of a jerk

Because what’s at play is:

• The amount of discomfort the “bad” thing creates for us (Musk = bad; Tesla = Musk; therefore Tesla = bad)
• The amount of discomfort we think action would create for us

Anything that decreases our discomfort, by definition, decreases the likelihood that we’ll take action.

The subtlety we discussed last week is that good diagnosis can decrease discomfort (“I figured something out!”), so it runs the risk of decreasing pressure to act.

And, remember, powerlessness is not a viable excuse because we all have some agency. Our problem is that agency involves inconvenience, discomfort, or personal / professional risk, and none of those is particularly pleasant.

Here’s an easy way to see what’s going on: for every observation we make, let’s add an action in the form of an “I” statement. As in:

“Coming into the office more doesn’t make sense because all we do is take our calls from the office instead of from home…and I’m planning to take forward a proposal to create a scheduled lunch hour for everyone twice a week.”

“I would love to advance in this company, but I can’t because I never get any feedback. I’m terrified, but I’m going to ask my boss for detailed feedback in our next 1:1.”

“People on my team just don’t take responsibility for their actions.  There’s no follow-through. I’m going to start a shared accountability chart and put my name and weekly To Do’s at the top of the list, and ask others to also fill it out each week.”

“My kids are addicted to their phones; it’s impossible to get their attention. We’re banning phones at mealtime and starting a ‘all phones in the kitchen drawer starting at 9pm’ house rule. These rules also apply to grown-ups.”

“The culture around here doesn’t make people feel valued for their unique contributions. I’m initiating a 10 minute ‘amazingness hack-a-thon’ each Friday where our team shares at least 1 amazing thing each team member did this week in or outside of work.”

“I just can’t get this build right because the user requirements aren’t clear. I’m locking the Product Manager, the subject matter expert, and my engineering lead in a room for 2 hours and we’re going to leave with 100% clarity on the spec.”

“Morale just isn’t what it used to be. People aren’t feeling a sense of connection. I’m going to schedule one 15 minute virtual coffee a week with a colleague, and I’ll come in with three questions that ensure I have a chance to learn more about them and their work.”

“We’re just not taking advantage of all that AI has to offer. Everyone is being so timid. I’m going to invite 10 people to a 2 hour after-work AI hackathon where we come up with 10 ideas that could move the needle, and we will start working on them.”

We can chuckle at the guy with the bumper sticker, but we’re all wearing bumper stickers of one kind or another.

The most dangerous one is the one that says, “I’m so good at figuring out what needs to be fixed that I’m able to stay completely in my comfort zone.”

All meaningful change involves some degree of discomfort and risk. How much is up to you.

It’s easy to think that learning new skills is about determination and willpower. Some people have it, and some people don’t.

While that is true, it is also incomplete.

Learning a new skill is a commitment to consistently spend time doing something poorly, and to refuse to give up despite how hard that feels.

In this way, it is as much about being willing to spend time in discomfort—physical or psychological—with little to show for your efforts, potentially for long periods of time.

The thing I’m currently bad at is the guitar.

I’ve been bad at it for about a year and a half now. Before being bad at it, I was nothing at it, and so bad is a big improvement.

In the beginning, my fingertips constantly hurt from pressing on steel strings—so in addition to not being able to play much of anything, I was in pain. Plus, nearly every note I played buzzed. That alone was a good reason to stop.

Then I learned a few chords, and quickly discovered that many of the most “basic” ones, including the F chord, were bar chords, requiring pressing down HARD on multiple strings with one finger. That was nearly impossible for me, a true beginner, and I couldn’t do it properly until about two months ago. So, 15 months of not being able to play a C-F-G-C sequence, which is about as basic a chord progression as exists. That was another good reason to stop.

Now I can play some songs, but playing a chord shift with a bar chord—G-major to B-flat minor, for example—is a 50/50 proposition at best. So, I’m practicing it. How? By repeating G-major—B-flat-minor hundreds of times. On a given night, I might play that 100 times terribly, and another 100 times less terribly. I’m not playing most of the song I’m working on (Summer of 69, randomly), I’m just playing those two chords for the better part of a practice session. By the end, it’s a little better, and I’m a little bit encouraged. And then the next day, it feels like I’m back where I started. And that’s frustrating too, and another good reason to stop.

The temptation to stop, you see, that is really the hard bit.

You have this idea in your head about what “good” will look like, and you’re so clearly far away from that ideal, that it can seem hopeless, and you can think, “Maybe this isn’t a good plan after all. Maybe I’ll never get there. Maybe I should stick to the things I already do well.”

The thing I notice, when I get back to the G-major to B-flat, is that it takes me fewer tries today to get to decent than it took me yesterday. I also notice that, after working on that transition, my (formerly) dreaded F-chord feels almost easy. Noticing the progress I’ve made, however small, is much more motivating than dwelling on the gap between me and everyone who plays the guitar so effortlessly.

So I dive in again.

Going towards the frustration.

Feeling that vulnerability.

Progressing much more slowly than I’d like.

But smiling at the voice saying “this isn’t all that fun, and it doesn’t sound that good. Maybe I’ll never get there” and just doing it again.

I’m talking about guitar but you can see that the discomfort is the same across disciplines: G-to B-flat could be learning how to integrate AI into your work, or keeping at running every day even though it’s never easy, or eating differently, or learning a new language, or deciding to take on virtually anything at all that you’re not great at today.

It’s not all that bad to be bad at something.

At least you’re doing the something.

And whatever you do, with intention and effort, for a long time, you improve at.

No matter what.

All thorny problems are difficult to solve. That’s what makes them thorny, after all.

But all good solutions can be explained simply.

That’s because all good solutions are hypotheses and nothing more.

Hence the simple 3-part explanation.

1. “Here’s what I believe lies at the heart of this problem.”
2. “Here’s what I propose that will address that issue.”
3. “Here’s why I think it will address that issue.”

Your goal is not to be right—in fact, the quest for rightness in the face of  complexity can be paralyzing, and inaction has its own costs.

Rather, your goal is to state these three points with utmost clarity.

This way, even if you don’t figure out the perfect solution (yet), you will at least know which of your three statements was right and which was wrong. Then you can iterate.

And if you can’t (yet) explain your problem and solution this simply, keep at it. Without this kind of clarity, it’s too soon to jump to implementation.

As the weather gets warmer, I start running and swimming more.

I’d love to track my runs and my swims, but I don’t have an Apple Watch or a Garmin.

Having more access to texts, news and Slack notifications—and even knowing more about everything outside my workouts—seems like it would be a bad thing, and I know if I owned the watch I’d wear it all the time.

More features = worse outcomes.

Similarly, I love my Kindle because it only does one thing well.

My favorite restaurants have small menus where everything is great.

Atoms shoes are incredible because they’re more comfortable than any other, and they reject design nonsense

In a world in which AI can do most things, being pretty good at a long feature set will be worse than saying “I will do fewer things exceptionally and reject all the rest.”

This requires two things:

1. Knowing your clients so well that you truly know what matters most to them
2. The bravery to leave everything but those features on the cutting room floor

Fabian has it right. What more do you really need to know?

My daughter played in a squash tournament last weekend.

In squash, each player has a computer-generated ranking that is reasonably accurate. It’s very easy to think that these numbers tell the whole story.

So, when my daughter lost her match (11-9, 12-10, 12-10 ) to a girl with a 0.10 lower rating, she was very upset.

The next morning, gearing up for day 2 of the tournament, I asked her a question: “If you had won 11-9, 12-10, 12-10 instead of lost, would you have been pretty happy with the result?”

“Yes,” she agreed, she would have.

We did the math together and noticed that the difference between what happened and what might have happened was 6 points.

6 points out of 64 were the difference between “existential crisis” (“Maybe I’m not really improving. Maybe I’ll never improve.”) and “I’m on the right track.”

There are situations in which the difference between winning and losing really matter: if your business runs out of cash but has a fundamentally sound business model, what matters is the cash. And if you “almost” hit your targets every time, then you might have a target-setting or an accountability problem.

But most of the time, we act like my daughter, allowing the space between our narrative of wins and losses to be much bigger than what actually happened.

Drawing the lucky last card feels like our just reward for playing the hand correctly. But the two events are, in fact, unrelated.

 

Two weeks ago, I was sitting with my daughter, helping her study for a test on linear equations. She’s in 8th grade, and we’re already getting to the point where my recollection of some of the math she’s studying is rusty. Soon she’ll be in high school and I’ll be of no use at all.

She had a set of problems that stumped both of us, and I told her to use ChatGPT to get an explanation for how to solve them so she could learn and practice the approach before school the next day. She told me she didn’t want to, that she didn’t like using AI, and that she’d just ask her teacher in office hours the next day.

While I have sympathy for her reaction, and a preference for her always talking to her teacher when she wants to go deeper, her approach isn’t going to work. The power and leverage created by AI tools is too much to pass on.

Just last month, a member of our 60 Decibels team created a working prototype to replace a piece of software that we’re paying tens of thousands of dollars for. The prototype took 15 minutes — 15 minutes! — to develop on v0.dev, and we expect it to replace the expensive, paid software we’re using sometime this month.

There’s a reason the Shopify CEO’s leaked memo — in which Step 1 reads “Hire an AI before you hire a human” — went viral. If we’re not retooling how we are running our organizations, we are already falling behind. Whereas if we’re diving in, we have the opportunity for unparalleled leverage.

Our job, then, is to keep on talking to the human—in my daughter’s case, her math teacher. But we need to go into that meeting having practiced and learned and honed our skill with the free tutor who now lives in each of our browsers. We need to take every repetitive task, every task that can be easily described, every part of the work that’s not uniquely leveraging our specific skills, relationships and insights, and find a way to have an AI tool improve or take over that part of the work.

 

It’s not as hard as it looks if we ground our apology in facts.

We say the following:

It is a fact that this happened.

I understand that this happened.

I apologize to you for my role in making this happen / not stopping it from happening. 

I feel genuine remorse about the fact that it happened.

Owning our mistakes is one of the bravest, most empathetic things we can do. No fooling.