Danny Halarewich
Menu

I build products and businesses from zero to scale.

I co-founded LemonStand and sold it to Mailchimp, where I went on to lead commerce. I later spent five years as CEO of Orbit, part of publicly traded Tiny. Today I’m building AI products and exploring what to take on next.

Get in touch See my work

Founded
LemonStand, sold to Mailchimp
Ran
Orbit as CEO, under Tiny
Now
Building AI products

Selected work

Building and running software businesses since 2008, as a founder, an executive, and a CEO.

Building now

AI coaching for trail and ultra runners.

Ontrack

I built Ontrack, from idea to working beta, to scratch my own itch. A training plan says what to do and Strava records what you did, but neither tells you whether the training is working. Ontrack compares the two, week by week, to show whether the work is paying off and you’re on track for your goal.

The plan comes from deterministic coaching rules, so every program follows proven training methodology. An AI coach handles the exceptions and adjusts the plan. It also supports you through conversation, on the things your watch can’t pick up.

Week 14 of 24

Phase

Peak

Avg compliance

89%

W1W6W12W18W24
km · distance
982
m · vert
14,180
workouts
62/70

Black Bear 50K · 10 weeks to go

Illustration Ontrack shows the whole training block at once, colored by whether each week hit its targets, not just the current week.
Explore the project

How I think about building

AI made building cheap, but it didn’t change what makes products win.

  1. 01

    Judgment is the job now.

    AI made producing work cheap: code, copy, prototypes, analysis. What’s scarce is knowing what’s good, which bet to take, and what to cut. I own every decision I put in front of a team, including anything an agent helped produce. I have to understand it well enough to question the reasoning and say “this is wrong,” with a why.

  2. 02

    Start with the problem, then choose the market deliberately.

    The reliable path begins with a real, painful problem, not a clever solution hunting for one. The market is a choice worth making on purpose: who exactly, and why them. For a startup it’s existential. In an existing company the market is usually already set, so the discipline moves to which problems you solve inside it. Retention is still the truest read on whether the problem was real.

  3. 03

    Stay close to the product and its customers.

    Use the product yourself, then watch customers use it on their own: they’ll struggle with things you’ve learned to work around. Support conversations explain what’s behind the numbers. Complaints mean they care enough to complain; silence is the red flag.

  4. 04

    Make the next bet testable.

    Every bet has one assumption everything else rests on. Name it, say what result would change the plan, and get the smallest possible test in front of customers. A forty-page PRD can still leave that assumption untested.

  5. 05

    Build less, and make what you build count.

    Every feature is a long-term relationship: you maintain it, support it, and end up defending it. A product needs a clear reason for the right customer to choose it. That guides what makes the cut and which details are worth another day on. Pick a few things and get them very right.

  6. 06

    Define good with real examples.

    For non-deterministic AI products, the eval is the spec: concrete examples of great and broken, a way to surface the failures that matter, and proof that a change actually helped. For deterministic work, plain acceptance criteria will do. Either way, check the results against what customers actually experience. The score is not the goal.

  7. 07

    Design the business while you design the product.

    A great product with no distribution or broken economics is a hobby, not a business. How customers find you should shape what you build, so choose the route early. Products that spread by word of mouth need instant value and a reason to invite. Products sold through a sales team look completely different. Price, cost to serve, and the path to the customer have to work by design. AI made serving every user expensive, so growth has to compound through retention.

  8. 08

    Small teams, full context, high agency.

    The best products come from small teams iterating fast, close to the customer. Give the team the business context, including what the work has to earn, so they can decide without waiting for me. I stay in the details with them. Judge the work by outcomes: do customers keep using it, and does the math work?

Writing

All writing

Contact

Let’s talk about what you’re building.

I want to work on something I believe in and take responsibility for making it succeed. I’m especially interested in AI products. If you need someone who can help shape the direction and build with you, I’d like to talk, even if there isn’t a defined role yet.

Or use the contact form