The Friction Audit: Mapping the Gaps Blocking Your Flow

How to find where your workflow is actually breaking down — before you hand off more to AI

When I first got serious about fitness, I got disciplined about one thing before I ever touched a weight: the night before. Workout clothes laid out on the edge of the bed. Protein powder set on the counter. Some nights I’d even shower before bed, so I wasn’t losing ten minutes to it in the morning.

None of that made me stronger. It didn’t improve a single rep. What it did was remove every small decision between waking up and being at the gym — so the only thing left to do at 6 a.m. was show up. That’s friction removal in its purest form: not trying harder in the moment, but making sure the moment requires less from you in the first place.

That’s the piece I want to bring into how we work with AI this week — different territory than last week. Last week was about what happens when you are checking — and the thing you’re checking against changes or disappears before you’re done. That’s still true, and it’s still yours to carry. But today isn’t about verification at all. It’s about something that happens before AI ever touches the task — and it’s easy to miss when you’re focused on what AI produces rather than how you set it up to produce it.

Direct answer: Friction isn’t the hard part of a task — it’s everything you have to decide or set up before you can even start the hard part. Reduce that, and the task gets easier without you doing it faster or trying harder. With AI, that means deciding your format, your throughline, your standards once — ahead of time — instead of re-explaining them every time you sit down to work.

Key Takeaways

Friction isn’t the difficulty of the task itself — it’s the setup cost you pay before you even begin. Reduce the setup, and the task gets easier, not just faster.
Laying out gym clothes the night before doesn’t make the workout easier — it removes the decisions standing between you and starting.
The same move works with AI, but it isn’t automatic — it takes intentional design up front, not a magic prompt.
I built that design into one place for my podcast newsletter: the throughline, the title format, the SEO structure are all decided in advance, so turning a transcript into a finished newsletter went from a couple of hours to about fifteen minutes.
Reducing friction isn’t just a productivity move for a solopreneur — it’s a nervous-system one. Every decision you don’t have to make live is capacity you get to keep.

The Problem

Most of us treat friction like it lives inside the task — this report is hard to write, this newsletter takes forever, this transcript is a slog. So we look for ways to do the task faster: a better prompt, a faster tool, more hours.

But look at the gym example again. The workout itself wasn’t the friction. Deciding what to wear, finding clean gym clothes, remembering the protein powder, showering with no time left — that was the friction, and it all happened before the actual workout. Once I moved those decisions to the night before, the workout itself didn’t change at all. What changed was how much stood between me and starting it.

Most people’s AI workflows have the same hidden cost. You open a blank chat, and before you get to the actual writing, you’re re-explaining your tone, your format, your audience, what “good” looks like for this piece — every single time. That re-explaining is the gym clothes you didn’t lay out the night before. It’s not the work. It’s what you have to do before the work can start, and if you’re doing it fresh every time, you’re paying that cost over and over for no reason.

I want to be clear this isn’t magic, and it isn’t automatic. AI won’t organize itself around your standards on its own — that takes the same kind of intentional setup as laying out clothes the night before. You have to actually decide, in a calm moment, what you want, and build that into the system. Skip that step and you’re just moving the friction from “the workout” to “the prompt,” which isn’t a fix at all.

This is also where it’s worth being precise about what kind of AI use we’re talking about. This is what I mean by agentic AI — not a single question answered once, but a system you’ve set up to carry a whole recurring task, start to finish, using standards you defined ahead of time rather than re-stating live. That’s a meaningfully different relationship than typing a fresh prompt every time, and it’s the one that actually removes friction instead of just moving it around.

The Evidence

You don’t have to take this on faith — the data backs up why setup, not effort, is the lever that actually moves things.

The UK Department for Science, Innovation and Technology’s AI Adoption Research, published January 28, 2026, found that among businesses currently using AI, 84% reported at least some human input or checking of AI-produced outputs, and 67% called that checking significant. But the detail underneath that stat is telling: most of that involvement wasn’t built into a repeatable process. It was reinvented, person by person, task by task — the AI equivalent of deciding what to wear every single morning instead of once the night before.

Anthropic’s own Measuring AI agent autonomy in practice (February 18, 2026) found that as people gained experience with agentic tools, their use of full auto-approve rose from about 20% (under 50 sessions) to over 40% (by 750 sessions) — but so did their intervention rate, from roughly 5% of turns to about 9%. Read together, that’s not people getting looser. It’s people getting more deliberate about where their setup needed reinforcing, based on what they’d already learned worked.

A preregistered field study of 758 BCG consultants found that on tasks inside AI’s tested capability frontier, GPT-4 users completed 12.2% more tasks, worked 25.1% faster, and produced work rated over 40% higher in quality — real gains, and none of them came from trying harder. They came from working inside a zone the researchers had already mapped in advance. Outside that frontier, without that mapping done ahead of time, AI users were 19 percentage points less likely to reach the correct answer than people working without it at all. The advantage wasn’t effort. It was preparation.

None of this says AI removes friction on its own. It says the businesses and individuals getting real gains are the ones who did the deciding in advance — the same move as clothes on the edge of the bed.

Design the Setup Once, Not the Task Every Time

Here’s where I actually use this, beyond the gym.

Every week, I release a podcast episode, and every Tuesday, a newsletter goes out built from that episode’s transcript. That used to take a couple of hours — reading the transcript, figuring out the angle, formatting it, checking the word count, making sure the title had both the throughline and enough SEO strength to actually get found.

None of that changed. What changed is when I do it. At the start of the month, I sit down with my content planner and decide the throughline for that month — the thread I want running through every piece. I’ve already defined the newsletter’s format: how a title should be built, how long sections run, what needs to show up where. I’ve trained a dedicated thread with all of that, once, so it isn’t a blank slate every Tuesday.

Now, when the transcript is ready, I hand it over with essentially one line — “you know the drill” — and it comes back already shaped to the format, on-theme, and SEO-considered. A couple of hours down to about fifteen minutes. Nothing about the actual writing got easier. I just stopped re-deciding the same things every single week.

That’s the same logic as a teleprompter for going live on video. The hard part of speaking on camera isn’t the speaking — it’s not knowing what to say next. Write the script once, load it into the teleprompter, and the hard part is already solved before you hit record. My podcast-newsletter thread is a teleprompter for my Tuesday writing. The throughline, the format, the SEO shape — all scripted in advance, so the actual moment is just execution.

And this is where intentionality matters most: none of this happens by accident. AI doesn’t know your throughline until you tell it, doesn’t know your format until you build it in, doesn’t know what “on-brand” means until you’ve decided that yourself, in a quiet moment, away from a Tuesday deadline. The friction only goes away because you did the deciding ahead of time — the same as choosing your workout clothes the night before, not scrambling for them at 6 a.m.

There’s a piece of this that goes beyond productivity. Every decision you don’t have to make live is nervous-system capacity you get to keep. As a solopreneur wearing every hat in the business, the cost of friction was never just time — it was the constant low hum of “what do I do first, what does this need to look like, did I forget something?” Removing friction, whether it’s gym clothes on the bed or a trained thread that knows the drill, isn’t just efficiency. It’s regulation. It’s one less thing asking for your nervous system’s attention before you’ve even started the actual work.

Put This to Work

Find one recurring task where you’re re-deciding the same things every time. A newsletter, a report, a social caption, a client update — something with a repeatable shape.
In a calm moment — not mid-deadline — decide the format, once. What does “done right” look like for this task? Length, tone, structure, throughline. Write it down.
Build that decision into the system, not just your memory. A saved prompt, a trained thread, a template doc — whatever means you don’t have to restate it live.
Test it on the next instance of that task and see how much of the setup cost actually disappeared. Adjust what didn’t land.
Notice where you’re still improvising live — that’s your next candidate for the same treatment.
When a task resists this — because it depends on a relationship, a judgment call, or something only you can actually vouch for — that’s a sign it isn’t a setup problem. That’s the line we drew a couple of weeks back, in The Judgment Split — where labor ends and something only you can vouch for begins.

FAQ

Isn’t building the format up front just extra work?
Once, for a task you repeat weekly, it pays for itself almost immediately. My podcast newsletter went from a couple of hours to fifteen minutes after one round of setup. If you’re not sure where else this kind of setup cost might be hiding in your business, the Streamline With Purpose™ Quiz audit maps it in about five minutes.

How is this different from just being more organized?
It’s the same instinct, applied deliberately to how you work with AI specifically. Being “organized” is a personality trait. This is a repeatable system — you build it once, and it holds the standard for you every time after that.

Does this work for tasks that change every time?
Less directly. This is for recurring tasks with a repeatable shape — the newsletter, the report, the weekly update. A true one-off still needs to be worked through live.

What’s the human advantage part of this?
Deciding what “done right” looks like — your throughline, your standards, your voice — is still entirely yours. AI can hold that decision and apply it consistently once you’ve made it. It can’t make it for you.

Why does this matter beyond productivity?
Because friction isn’t just a time cost — it’s a nervous-system cost. Every decision you don’t have to make live is one less thing pulling at your attention before the real work even starts.

The Close

I didn’t get more disciplined by trying harder every morning. I got disciplined by deciding the night before, so morning-me didn’t have to decide anything at all. The same principle runs my Tuesday newsletter, and it can run almost anything else you hand off regularly — not because the AI got smarter, but because you did the deciding once, on purpose, ahead of time.

As a solopreneur, reducing friction is one of the most powerful ways to regulate your system and maintain balance while still trying to achieve your passion.

As a solopreneur, reducing friction isn’t just a productivity move — it’s one of the most direct ways to protect your capacity to actually run the business you built. You don’t have to build every system alone.

Amplify You Club membership card — $144/month, with Shilpa Lewis on a video call with community members, offering monthly AI workshops and ongoing support for solopreneurs

Build your small business alone — or build it with expert support. $144/month gets you 2 live workshops, 1 Skill File or Playbook, 1 group coaching session, plus the Member Resource Vault and ongoing support.