4 Lessons from Six Months of Startup Life, After over a Decade in Big Tech

I braced for the grind. Six months in, that was the easy half.

Half a year ago I left a decade of engineering leadership to explore entrepreneurship.

I thought the hard part would be the grind. Longer hours, faster execution, nobody to hand work to. That part turned out to be the easy half. What didn't sort itself out was knowing whether the thing I was building was worth building, and getting anyone to believe me once it was.

Looking back, I don't regret the decision. Six months out here put me in front of an entirely different set of problems than another few years as a corporate leader would have, and the mistakes are the part worth writing down. The four below are judgment calls rather than moves, so they're useful whether you're weighing the jump yourself, sizing up an early-stage company from the outside, or already out here and comparing notes on what building looks like in the AI era. Most of them were expensive.

1. You can't tell if you picked the right problem from the inside

Almost everyone I talk to about starting something is stuck in one of two places:

• Still looking for the idea. They read, they research, they wait for a problem clean enough to justify the leap. The searching quietly becomes the activity, and nothing is ever quite it.

• In love with the idea. They're all in and can talk about it for an hour. Any suggestion that it might be the wrong problem gets heard as a failure of nerve, so they don't pivot. Pivoting would mean the conviction had only ever been enthusiasm.

This costs more than it used to. When a prototype took a quarter, choosing badly cost a quarter, so the choosing came wrapped in process: research, sizing, review. When a prototype takes a week, that process has nothing left to protect you from. The cost of building the wrong thing collapsed. The cost of not knowing whether it's the wrong thing went up, because it's now the only thing standing between you and a year spent well.

You also can't resolve it by thinking harder. Conviction and sunk cost produce the same feeling in the body.

Three things helped.

Ship before you're comfortable. Everyone says that if you feel comfortable when you ship, you shipped too late. I'd agreed with that line for years inside a company where agreeing with it cost me nothing. Living it is a different thing, and I now think it's the most important item on this list, because shipping early produces the outside signal you cannot generate from inside your own head.

Look for a design partner, not users. What I didn't have language for at the time: at that stage you want one person who has the problem and will sit in the mess with you, shaping the thing while it's still cheap to change. A rough version is what qualifies them. A polished one mostly tests whether people like your taste. An embarrassing one tests whether somebody will tolerate real friction to get the outcome.

Then get them to pay. A design partner is not the finish line, which took me longer to learn. A good one will co-design with you happily and indefinitely. They like you, the problem is real, the sessions are interesting, and none of that is evidence. The milestone that counts is converting a pilot into a paid contract, because that's the first moment somebody has to defend the decision with their own budget and their own name on it.

That conversion is also the only reliable way I've found to tell the two traps apart. You can't separate conviction from sunk cost by how it feels. You can separate it by whether anyone has paid.

2. Trust got scarce as distribution got hotter

Trust was always the thing that decided whether a stranger gave you a chance. That part didn't start with AI. What AI did was flood every channel at once, until the shortage became impossible to ignore.

Anyone can now produce a clean landing page, a personalized cold email, a demo video, a plausible case study. So every channel that used to work is crowded with things that look competent. The customer's problem stopped being finding options and became telling them apart.

Trust is the filter they use.

For months I thought I was bad at marketing, because I kept rewriting the landing page, the messaging, the follow-up. What I was doing was answering one question over and over:

Why should I trust you?

Once I saw that, the scattered work turned out to be a single piece of work. Growth is the business of lowering the cost of trust. There are only so many mechanisms that do it:

• Community — people who already trust each other, deciding together. Reddit is the clearest version. The platform punishes marketing and rewards history, so an account with years of useful comments outperforms a funded campaign, and there's no way to buy that or start it the week you need it.

• Referrals — borrowed trust from someone who already paid the cost.

• Enterprise champions — one person inside the org staking their own credibility on you.

• Brand — trust accumulated in public, slowly.

• Product quality — trust earned per use.

• Safety and reliability — trust that survives the first failure.

• Customer success — trust maintained after the sale, which is where most of it gets lost.

Notice what they have in common. AI can write your outreach. It cannot be your reference customer, and it cannot sit in the room when a VP puts their name on your pilot. Every mechanism on that list is paid for in time, and time is the one input that didn't get cheaper.

That's why I think this is where the durable advantage sits now. It's the part of the stack that didn't get commoditized.

3. Don't ask for behavior change at the start

The instinct is to sell the better way of working. It's the wrong opening move.

Our pilot went smoothly. The product worked, the people using it liked it, the sessions were good. What didn't go smoothly was the conversion to paid.

For a long time we read that as a solution problem. Sharpen the product, close the remaining gaps, make it a little more indispensable, and we'd clear the paid milestone. So we kept optimizing.

That wasn't it. What sat between the pilot and a contract had almost nothing to do with how good the thing was. Getting to paid required someone to change how their team worked, find money that wasn't in a budget, and defend the choice to their boss. Each of those is a tax, and all of them get paid before anyone has evidence you're worth it. No amount of polish reduces them.

The version that worked started from the opposite end. Fit the workflow people already have. Sit inside the budget that already exists. Ask for as close to zero change as possible, and let the product earn trust by being useful in the position it already occupies.

Behavior change is not off the table. It isn't something you get to ask for on day one. Once you're inside the workflow and the trust is real, moving people costs them far less, and they'll often propose the change themselves. Sequencing is the whole lesson: fit first, influence later.

4. "Why me?" is worth answering even if nobody asks

"Why you?" is the classic investor question. What makes you the right person to solve this particular problem? I think it's worth sitting with whether or not you're raising money, because it's the only question in the whole process that a framework can't answer for you.

It breaks into two smaller ones.

Do you see something here that other people don't? Not a better execution plan, and not more enthusiasm. An asymmetry: something your particular history put in front of you that isn't obvious from the outside. If the honest answer is that you'd be running the same play as anyone else with a similar résumé, that's worth knowing before you commit years to it.

Would you regret not taking the shot? There's no analysis behind this one. It's what's left after the market sizing and the competitive scan are done, and for most people I've talked to it's the question driving the decision anyway.

I reached a point where I could see the shape of a viable company. I had the background to build it. The second question settled it: I didn't want to spend the next five years thinking about that problem. No data behind that. It was still the right call.

Annie Dillard wrote that how we spend our days is how we spend our lives. The startup version of that is harder to look away from. The problem you pick becomes the thing you think about on walks, in the shower, at two in the morning. It becomes the people you talk to most, and the questions you turn over when you meant to be resting.

A framework can tell you whether a market is attractive. It can't tell you whether you want to spend your mornings there.

That one stays yours.

I'm keeping a record of this transition as I live it, subscribe if you'd like the rest of the journey, and follow me on LinkedIn where I post the shorter version of what I'm learning.