Shipping HarkDog — a vertical SaaS CRM for professional dog trainers

The thesis, in one paragraph

HoneyBook is a $45/month CRM that charges every service business in America the same price for the same generic feature set. Underneath those businesses are dozens of verticals — photographers, trainers, florists, cleaners — each with their own vocabulary, workflow, and client journey. The generic tool wins on distribution. The vertical tool wins on fit. The gap between “this is a form field” and “this is a milestone in my puppy program” is where the vertical SaaS lives.

HarkDog is that bet for professional dog trainers. The nouns are the ones a trainer already uses — clients, dogs, programs, sessions, pathways. Not contacts, projects, tasks. Everything the trainer does with a client has a dedicated shape, and the client sees the same shape from the other side.

What actually shipped

1. A trainer dashboard

Lives at app.harkdog.com. Clients, dogs, projects, sessions, contracts, invoices, messages, programs, products, reports, templates, automations, team seats. Pathway templates — multi-session training programs with milestones, trainer objectives, and owner homework — because that’s how trainers actually structure a 6-week puppy course. Graduation triggers automatically when the last milestone lands.

2. A client PWA, same URL

Lives at app.harkdog.com/c. Installable to the home screen with a manifest and service worker. Push notifications when the trainer sends a message, uploads homework, invoices, or marks a milestone complete — VAPID keys set up and keyed per client. Dog owners get the lightweight end of the app without needing an App Store listing.

3. A public trainer profile on their own subdomain

Each trainer gets <slug>.hark.dog — a public landing page with their programs, testimonials, and direct Stripe-Checkout booking. The hark.dog nameservers are delegated to Vercel so wildcard SSL for every *.hark.dog subdomain is auto-provisioned. New trainer signs up, their subdomain works five minutes later.

4. Stripe Connect, pass-through only

HarkDog takes 0% on client charges. When a dog owner pays the trainer for a program, the money goes to the trainer’s connected Stripe account directly — HarkDog never touches it. The only revenue HarkDog takes is the subscription. That matters because trainers who are already running six-figure businesses don’t want a 3% tax on everything they already charge. $19 Starter / $49 Pro ($39 annual) / $89 Studio ($79 annual) — prices chosen to sit alongside HoneyBook’s pricing, not undercut it.

5. A marketing site at hark.dog

Separate Next.js app — headline, hero, real product screenshots (not mockups), pricing, feature pages. “Less admin. More training. More growth.” is the leading line. The product screenshot below the hero is pulled straight from the live app — because a prospect clicking through should see what they’re actually buying, not an illustrator’s version of it.

How it’s architected

A Turborepo monorepo with two Next.js apps — apps/web for marketing and apps/app for the dashboard + client PWA — sharing three packages:

• · packages/db — Prisma schema, pinned to 5.19.1, generates on every build
• · packages/lib — shared helpers, server-only modules at subpaths (push, SMS, Google Calendar, receipts) so the client bundle never pulls them in
• · packages/ui — shared components across both apps

Database is Neon (Postgres). Storage is Vercel Blob for client-uploaded images, fonts, and attachments. Transactional email is Resend from a verified sender domain. Push is standards-compliant Web Push via VAPID. Both apps deploy on Vercel under the same team — hark for marketing, hark-app for the product.

The first trainer in production

A professional dog trainer in Houston is the first operator running her business on HarkDog in production — actual clients, actual programs, actual money moving through Stripe Connect. Having a real trainer in the system means the rough edges show up now instead of after launch.

That’s the right shape for a vertical SaaS v1 — one real operator stress-testing the product before I open the gates to the next cohort.

Why vertical, not horizontal

Horizontal SaaS — the HoneyBook or Dubsado shape — wins by being “good enough for everyone.” That’s a defensible model when you have a hundred-million-dollar sales budget. For a one-person shop, “good enough for everyone” is a losing position.

Vertical SaaS wins because the product can speak the operator’s language. A dog trainer who logs in to HarkDog sees pathway progress, not project status. A paving contractor who logs in to Modern Drive sees drive length / drive width / drive thickness, not line item 1, line item 2, line item 3. The friction goes away. The “does this tool get me” moment arrives in minute one.

Both of these products are the same bet made twice.

What’s next

Open the gates past the first trainer. Line up trainers two through ten through outbound, not inbound — because the first ten users of a vertical SaaS tool should be real operators I can text with, not sign-ups off a Facebook ad. Keep writing publicly about what shipped and what broke.

If you train dogs for a living and this is pointing at a problem you’ve been working around with spreadsheets, HoneyBook, Google Calendar, and a group text — the trial is 14 days, no card required.