Case studies
Real systems, inside real companies.
What we've built, and what we can build for you.
Case study · Hulken
Hulken. A $50M+ consumer brand, run by a team of 7.

Revenue
$50M+
Team
7
Trajectory
$100M path
System
Multi-agent OS
The challenge
Running a consumer brand at $50M+ with seven people is not a staffing decision. It is a coordination problem. Inventory signals lived in one system, marketing performance in another, decisions in chat threads that vanished.
What we built
A multi-agent operating system connecting product, inventory, marketing, and operations into one coordinated layer, with the underlying data made trustworthy first. AI sits on top, reading and writing back, so every action shows up where the team already works.
The result
Decisions in hours instead of weeks. The team stays seven people while the company scales toward $100M.
Origin
The results were transformative enough that Hulken's leadership didn't just keep the system. Alex Schinasi (Hulken's co-founder and President) and Lee Rotenberg (Hulken's CRO) joined forces with Or Amir, who built it, to found Northstone and bring the same work to other companies.
Before this, growing meant hiring - every new bit of complexity needed another person to manage it. Northstone flipped that. They didn't just add AI on top of our mess. They rebuilt how information actually moves through the company and then put agents on top of it. Now seven of us run a large business and make decisions much faster. I genuinely can't imagine going back.
See something you recognize in your own company?
Case study · WHIP
One operating system for a studio that runs across four cities.

How WHIP replaced six disconnected tools with a single system, and put AI on top of it.
Client
WHIP, a remote creative studio: branding, design, and visual production.
Shape
A fully remote studio working across three time zones
Before
Work scattered across a task tool, two file drives, chat, and separate invoicing
Engagement
Operating-system design plus a layer of AI agents
The challenge
WHIP does premium brand and production work across time zones. The creative was excellent. The operation around it hadn't kept pace. Project volume scaled fast and the administrative weight compounded: scattered task tools, multiple file drives, disjointed chat channels, and decisions made on calls that were hard to track. As the studio grew, the coordination cost grew with it. They did not need another tool. They had plenty. They needed one system, built around the studio's creative rhythm, that the whole studio could run on.
What we built
We started where the work hurt and built the foundation first: a single source of truth, structured around the way the studio actually works. Every project is assembled from standardized building blocks, brand identity, web design, photoshoot production, and more, each carrying its own templated milestones, approvals, and payment steps, over 150 in all. One connected place for clients, projects, deliverables, and contracts.
- Personal AI assistants for every team member, handling the daily overhead: calendars and scheduling, email triage, action items, meeting notes, and a morning brief across active projects.
- A project intake system that turns a short conversation, or a single form, into a complete project plan: a schedule worked back from the delivery date, parallel workstreams, approval gates, and a clean client-facing timeline on demand.
- A client portal that mirrors the studio's system in real time: projects, deliverables, status, and schedules in one shared hub, where clients leave comments that flow straight back into the studio's workspace. Nobody emails to ask where things stand.
- A routing layer that moves work between these systems automatically, with an approval gate that encodes the studio's own rule: nothing reaches a client until the creative director signs off.
The result
The studio now runs on one operating system instead of a pile of disconnected tools. Every project follows the same repeatable path from intake to delivery, with the schedule, approvals, and client visibility built in. Contracts that used to take real time now generate in minutes. Team members start the day with an automated brief instead of chasing status, and new hires are onboarded straight onto the system, learning one place instead of ten, while clients follow progress in a live portal instead of email threads. And because it was built for handoff, the studio owns and runs it themselves.
Honestly, it's the first time the whole studio runs on one system, built around how we actually work. Nothing slips through the cracks anymore, and we get to spend our time on the creative.
What we can do.
A few short examples showing the range.
Operations team, buried in manual work.
We turned scattered spreadsheets into one source of truth and automated intake, routing, and reporting, giving hours back every week.
Regulated enterprise division.
An engineered, auditable system with a human in the loop, so AI drafts and a person always decides.
Fast-scaling startup.
Quick wins in week one on the obvious manual pain, then an operating layer built underneath as they grew.
More ways to work with us.
Not every company starts with a build.
AI training for your team
Two focused hours with your team, live at your offsite or over a call. We take people from AI-curious to using it on real work: your tools, your use cases, hands on keyboards. Teams leave with working habits, not theory. Often the first step before a bigger engagement.
Format A 2-hour working session, on-site or remote
Best for Teams that want practical AI fluency, fast
AI strategy advisory
For organizations that have the resources to build but no clear path to results: platforms deployed, models licensed, and adoption stuck. We bring the operating-system playbook: what to build, in what order, who owns it, and how to make it stick.
Format An advisory engagement, scoped to your organization
Best for Leaders with in-house teams whose AI investment isn't landing
