A closer look at how we build marketplaces.
A closer look at how we actually work: the three phases every marketplace project goes through, where AI fits in, and where human judgment takes over.
The full process, from any starting point.
You don't have to start at the beginning. These phases represent our full process, but your starting point depends on where you are right now. One experienced team, no handoffs, however you engage with us.
Discovery and economic pressure-test
We sit with you for a two-day intensive. Unit economics, take-rate sensitivity, supply concentration risk. We tell you which of your assumptions will not survive contact with users.
Wedge selection
Most marketplaces fail because they tried to be a marketplace too early. We help you pick the smallest wedge: one geography, one category, one supplier type. The one that proves the loop.
Validation prototype
A clickable, hi-fi prototype you can put in front of ten target users in week three. Not Figma slides. A real, navigable thing that surfaces the questions Figma cannot.
Supply story
Cold-start strategy with named first-twenty supply targets. We have done this manually, by hand, for our own marketplaces. We will do it with you for yours.
Architecture and design system
Senior decisions on stack, data model, integration boundaries. Design system locked in week one so the build does not drift. AI does not get to make these calls.
Core flows, AI-augmented
Listing creation, search, booking, messaging, payments. AI scaffolds the boilerplate, generates the repetitive UI, drafts test coverage. Engineers review and ship every line.
Trust, money, and data. By hand.
Anything that touches identity, payouts, disputes, or user data is written by senior engineers. No AI shortcuts. No exceptions.
Weekly demos with real users
You see working software every Friday. Three target users see it every other week. The feedback loop is the project plan, not a Gantt chart.
Pre-launch hardening
Performance, accessibility, security review, payments edge cases. The boring work that determines whether your launch holds up to its first hundred users.
Soft launch with seed supply
You go live with the first twenty listings already loaded, the first ten transactions already lined up. We do not launch to silence.
Cohort health and retention loops
We watch the second-transaction rate, the supply churn rate, the GMV-per-cohort curve. We tell you what to fix in priority order, and ship the fixes.
Supply loyalty and referral mechanics
The plays that turn casual suppliers into your best growth channel. We have run them on our own marketplaces. They work, but only if you commit.
Speed is what is new. Senior thinking is not negotiable.
We get the AI question on every call. Here is the honest answer.
What AI does in our process
- Generates boilerplate and infrastructure code: auth scaffolds, CRUD endpoints, deployment configs.
- Scaffolds standard CRUD and admin UIs: list views, edit forms, table layouts.
- Drafts test coverage: unit tests, integration tests, edge-case enumeration.
- Accelerates documentation: API references, internal handoff notes, changelog drafts.
- Handles repetitive integration work: webhook plumbing, schema mapping, format adapters.
What humans always do
- Architecture decisions: data model, integration boundaries, scaling shape.
- Marketplace-specific judgment calls: take-rate, supply incentives, trust mechanics.
- Code review on every line that ships. No exceptions, no AI-only commits.
- Founder-facing strategy: what to build, what to cut, when to launch.
- Anything involving trust, money, or user data. Written by hand, every time.
Four things we promise, every engagement.
You own the GitHub repo from day one.
Not at handoff. Not after final invoice. From day one.
You own the Figma file from day one.
Design files, prototypes, source of truth. Yours.
A named senior practitioner.
Every project has one. They are on your call list, not behind a project manager.
60-day bug guarantee.
Anything we ship, we fix free for sixty days post-launch. No fine print.
Sharetribe primary, custom where it matters.
We start with Sharetribe because it solves the marketplace problems most platforms re-solve badly. We add custom development when your wedge needs it.
What working with us actually looks like, week by week.
This is a ten-week Build & Launch engagement. Grow and Fractional rhythms differ, but the pattern is the same.