The Unlock Isn't a Better Prompt
- Mar 9
- 5 min read
How your setup matters more than your skills
In the fall of 2000, my MBA classmate Rob Page and I flew to Israel for our international field study. We were working on a project for Cisco Systems, and the assignment was straightforward: figure out what the killer app for broadband would be. Cisco's hypothesis was simple — if the right application emerged to drive consumer broadband adoption, demand for their networking hardware would follow. Our class split up and fanned out across the world. Rob and I got Israel.
It was a good place to look. Israel's mandatory military service had produced a generation of engineers who understood radar, wireless systems, and signal processing at a level most Silicon Valley startups couldn't match. The startups we met were technically serious and thinking well beyond the PC as the primary connected device.
We came back without a killer app. What we came back with was a thesis: it probably wasn't going to be one application that drove broadband adoption. It was going to be frictionless connectivity — plug and play setup, devices beyond the PC connected to the network. The internet had to stop being something you went to and start being something that was just there.
We were partially right.
We got the friction part right. What we didn't see was what happens to behavior when friction disappears entirely.
A Pew Research study published in 2002 captured it. Broadband users did about seven different things online on a typical day. Dial-up users did three. Broadband users were twice as likely to go online multiple times a day. The researchers called it "the always-on information appliance" — and they found that the presence of a broadband connection was the single largest predictor of how intensely someone used the internet. Not income. Not education. Not how long they'd been online. The connection itself.
Dial-up users batched. They accumulated a few things they needed to do — send an email, check a site — then connected, did them, and disconnected. Broadband users acted in the moment, because there was no longer any cost to doing so. Voice input does something similar — when you can think out loud instead of type, the threshold for what's worth asking drops further. If you missed my Saturday Morning Coffee on it, Voice-to-Text for AI: From Brain Dump to Structured Output is worth the eight minutes.
The internet had become ambient.
Cut to 2025. I heard about someone who kept Claude open on a dedicated second monitor — always visible, always present, never minimized. I thought it was interesting enough to try. So I adjusted my setup and did exactly that.
My usage went up. But that wasn't the interesting part. What changed was the category of problems I started bringing to it.
A few weeks in, I was in the middle of grading assignments for one of my university courses, nagged by a feeling that I wasn't being fully consistent. Because Claude was right there, I asked. It turned out I was letting writing style cloud my judgment on substance — giving higher marks to well-written answers that were actually thinner on content. I wouldn't have opened a browser tab for that. But I didn't have to. The question cost me nothing to ask, and the answer changed how I finished the grading session.
Around the same time, I was in the middle of trying to organize my business development activity and getting frustrated with my CRM. I kept second-guessing how to characterize where different prospects actually were. Seasoned salespeople have instincts for this. I don't. So right there, mid-workflow, Claude and I worked through a simple, structured set of prospect statuses that actually matched how I think about relationships. Again — not something I would have scheduled time for. But ambient, it just happened.
Neither of those is a product leadership task. That's not a coincidence. Ambient availability didn't just make me more efficient at the work I was already doing. It expanded the domain of problems I considered worth bringing to AI in the first place. The threshold for what's worth asking dropped — and that turns out to be the unlock.
"The threshold for what's worth asking dropped — and that turns out to be the unlock."
Most of the conversation about AI adoption focuses on skill. Learn to prompt better. Understand the models. Build workflows. That's all useful. But it assumes the limiting factor is capability, and for most people, it isn't. The limiting factor is the friction of access — the small but real cost of switching contexts, opening a tool, formulating a question formally enough to feel worth asking.
Remove that friction and behavior changes. Not because you got smarter about AI. Because you stopped having to decide to use it.
"Remove that friction and behavior changes. Not because you got smarter about AI. Because you stopped having to decide to use it."
I want to be clear about something. A second monitor and an always-open tab will change how you work. It won't build AI capability across your team. It won't create the workflow infrastructure that makes AI adoption stick organizationally. It won't solve the harder problem of helping a group of people develop new instincts together. Those are different challenges — and the ones that most leaders are actually trying to solve.
But this is where individual adoption starts. And most people haven't gotten here yet because they're still waiting for the killer app, still optimizing their prompts, still treating AI like a tool they go to rather than one that's simply there.
There's a next chapter here worth paying attention to. The broadband shift was about always-on availability. What's coming is always-on context — AI that isn't just present, but aware. Aware of what you're working on, what was discussed in your last meeting, what decision you're in the middle of making. When ambient availability combines with persistent context, the behavioral shift will be larger still.
The experiment itself is simple. Give AI a dedicated place in your physical environment. Not a tab buried under twelve others. A persistent presence. See what you reach for, and when.
The unlock probably isn't what you've been working on. And if you're already thinking about what this means for your team rather than just yourself — that's the more interesting question. You know where to find me.
Break a Pencil,
Michael
P.S. Hit reply and tell me what you end up doing with it that surprised you. I'm genuinely curious what problems show up once the friction is gone.
P.P.S. If this resonated, forward it to someone who's still optimizing their prompts and wondering why AI hasn't clicked yet. They might be one setup change away.
Comments