Overview
Atlas Pet Company makes premium dog leashes, harnesses and gear. Founder Sam Alter started the company about 11 years ago, after selling analytics software to some of the fastest-growing eCommerce brands of the time. Atlas has been bootstrapped from day one.
Everything happens in-house in Colorado. The team makes every product, ships every order and answers every support ticket. Atlas products carry a lifetime warranty, even if your dog chews through them, and the team repairs them for free. Atlas has won Wirecutter's best dog leash and Outside magazine's best dog harness.
Atlas runs all of this with four full-time employees, one part-timer and Sam, who owns everything that isn't production or shipping: marketing, advertising, content, creative, brand strategy and product.
This case study covers why Sam runs lean, the systems Atlas tried before Fulfil, why they chose Fulfil over NetSuite, and how an AI-pilled founder builds on top of his ERP instead of hiring around it.
Why Atlas Runs Lean
Sam never wanted a big company.
"Obviously mine's a little bit different because it's been bootstrapped since the beginning, and a lot of it was around wanting to own a hundred percent of the company and not build a huge company, but to have a small company that makes great products."
— Sam Alter, Founder
The first holiday season after setting up the warehouse made that philosophy concrete. Atlas hired four people to make dog leashes. Three were let go within a month. The fourth lasted another six.
"It was a really eye-opening exposure to quality control and how bad it is if you don't have people that care. Even though we were paying people around thirty dollars an hour, it didn't change their philosophy on how to care about what they're making. I'd rather have fewer employees who are getting paid better and getting more pride from the job they're doing."
— Sam Alter, Founder
Rather than add a middle manager, Sam focused on efficiency in pick and pack, in manufacturing, and in systems that let people do their jobs without someone managing them.
"That was the real wake-up call, way before AI was even an option. It led me to really focus on efficiencies in the warehouse, in our pick and pack, in our manufacturing, and generally on how to build systems that let people do the things they need to do without management being necessary."
— Sam Alter, Founder
The Systems That Didn't Work
Atlas came to Fulfil the hard way. Their previous ERP, Locate Inventory, was acquired and shut down, and customers were given eight to twelve months to move.
"We were using Locate Inventory, which got acquired and shut down. We were all told we had eight or twelve months to move. We implemented Odoo, we implemented ShipHero and Katana, and we implemented Fulfil. We stuck with Fulfil and didn't stick with the others."
— Sam Alter, Founder
Katana and ShipHero. Running manufacturing in one tool and shipping in another meant Atlas had to sell products to itself to get them into the shipping system.
"That's how we were doing it with Katana and ShipHero: make everything in one tool, sell it to the other business in the other tool, then receive it. It was so complicated from an accounting perspective, and it's not what we actually do in the process."
— Sam Alter, Founder
Odoo. Odoo was the only other all-in-one system on the list, but it needed so much upfront customization that Sam worried no future hire would know how to use it.
"It felt like we'd build a system nobody else would know how to use, and we'd never be able to hire anybody who'd used a better system than the one we threw together."
— Sam Alter, Founder
What Atlas wanted was an ERP that came with the processes already worked out.
"We wanted a tool that would tell us what these processes should be and how the biggest brands in the world do these things, because we didn't want to invent them at a million in revenue and hope they didn't break as we grew."
— Sam Alter, Founder
With manufacturing, inventory, shipping and accounting in one system, Atlas makes a product and ships it from the same records. No internal sales, no second receipt.
Fulfil vs. NetSuite: $248K and Nine Months
Atlas also got a quote from NetSuite.
"When we looked at Fulfil, we got a quote from NetSuite that was about ninety-eight thousand dollars, plus another hundred and fifty thousand in implementation over a nine-month process. We pay a lot less for Fulfil, but more importantly, we have direct access to the team when things go wrong."
— Sam Alter, Founder
For Sam, the bigger issue than price was who he'd be paying to make changes.
"I don't think I'd be able to invest as heavily if I went through a third party I was spending a hundred thousand dollars on and feeling bitter about every interaction, feeling like I have to keep paying them, and keep paying them, and keep paying them. Fulfil built the software, gave me MCP access, and lets me build on top of it without having to pay your team."
— Sam Alter, Founder
See related: Fulfil vs. NetSuite
Buying the Promise of Open APIs
Fulfil's API docs are only available to customers, so Atlas bought partly on trust. Sam credits Chantel, his Fulfil sales rep, for being straight about what he'd get.
"Fulfil is really focused on building open tools that let other tools connect to them. You do have to buy into that during the sales process, because you can't actually see the API docs before you're a Fulfil customer. I understand why now, but you're buying the promise of having access. I'm glad Chantel did not overpromise."
— Sam Alter, Founder
That promise turned out to be the foundation for everything Atlas has built since.
From a $9,000 Developer Project to Building It Himself
Two years ago, Sam needed to schedule production more precisely. Give operators too little work and it still takes a full week. To set realistic weekly targets, he needed cycle time data that Fulfil was already tracking, displayed in a way he could plan from.
He hired a developer. Six months and $9,000 later, he had nothing that worked, and three months until Black Friday prep. So he started building with AI himself, in Replit, before Claude Code existed.
The data was already there. Operators had been tracking production cycles in Fulfil from the start because Fulfil uses them to calculate product costs.
"I didn't have to train my employees to start tracking cycles, because that's how we were doing it already, since Fulfil needed it to track product costs. We already had all that data and the workflows in the shop. We just needed to use it and visualize it so that we could actually plan from it."
— Sam Alter, Founder
"The app I built had a working prototype in three weeks, because I was able to do it myself and not have to translate everything I needed to a developer, have that person build it or not build it, and then take another stab at it."
— Sam Alter, Founder
The Tools Atlas Built on Fulfil
That first project opened the door. Sam now builds tools on Fulfil's API that the team uses every day.
"The Fulfil APIs made everything possible. I didn't have to rely on the Fulfil front end, and I didn't have to rely on the Shopify front end."
— Sam Alter, Founder
- Production planner. Plans each week's production for every operator. It pulls cycle data from every tracked work order in Fulfil, calculates expected time per item, and compares it against actual time, including hours from Gusto that aren't linked to a cycle. It also shows when Atlas will run out of raw materials or finished goods by variant, using Fulfil's supplier lead times, purchasing data and MOQs.
- Repair portal. Every repair under the lifetime warranty needs a label in and a label back. When something can't be repaired, the team picks a new or blemished replacement and sends it on the original repair label or as a Fulfil shipment, creating a customer service sales order against the original order.
- Shopify order and customer blocks. Fulfil data shows on the Shopify order and customer pages: past order history, repairs, returns, shipments and support tickets.
- Operator iPad app. On the same iPads operators use for Fulfil cycle tracking, it shows shipments past their SLA, outstanding returns and support tickets, and items waiting on assignment with their open manufacturing orders. Operators can ask it questions too.
The team prototypes in micro apps with Claude Code, then builds the finished tool from there.
"We build a starting view with micro apps using Claude Code, refine it there, and then give it to Replit and say, 'Use the micro app we built here to build this into app XYZ.' I'd say micro apps work really well for on-demand dashboards, things where you just want a quick view and don't want to build it into other softwares."
— Sam Alter, Founder
Kai: The Ops Lead That's an Agent
Sam spent two and a half years trying to hire a head of ops. Three people took the role. None lasted 90 days.
"After the last person left in under three months, I got pretty fed up with spending three months training somebody who then goes nowhere. So I took their job description, dumped it into Claude Code, and said: 'This person's gone. We need to replace them with an agent, and we need to build this into Fulfil and Claude Code so it's all automated and scheduled.'"
— Sam Alter, Founder
The last hire left in March. Kai, named for Kaizen, was running by April, and has been running for about six months.
"It's been so much easier to invest in an agent that gets smarter and better over time, always does all the work, works all through the night, and doesn't ask whether he needs to bring his laptop home at the end of the day."
— Sam Alter, Founder
Kai works through Fulfil's MCP. Sam explains problems to it the same way he'd explain them to a new hire.
"I don't have to explain what I need beyond how I already explained it in a job description or a process doc. I just say, 'This is my problem, this is how we're doing it today, how can this get better?'"
— Sam Alter, Founder
See related: Implementing AI in your operations
Full Access, Not Guardrails
Most AI advice says to lock agents down. Sam does the opposite.
"There's a lot of talk about permissions and control in AI and keeping agents fairly restricted. I've taken the complete opposite approach. I have so much to gain from my AI agents being fully autonomous with full access, and it's almost never broken something it couldn't fix or roll back."
— Sam Alter, Founder
Here's what that looks like at Atlas:
- Support inbox. Kai monitors the support inboxes. When a customer emails at 1 a.m. to change a shipping address or a color, Kai holds the order, updates the address or items, and re-releases the shipment. If it can't make the change itself, it flags it and keeps the order on hold.
- Two-bin parts system. Atlas combined two parts racks into one and moved to a two-bin Kanban system. Through Fulfil's MCP, a Claude project updated every preferred warehouse location, bin quantity and case pack quantity over about three and a half weeks of background tasks, before the racking was even ordered.
- Accounting in Fulfil. Atlas had used Finaloop for three years because running Fulfil accounting would have meant a full-time bookkeeper. Last month they migrated everything into Fulfil. An agent now keeps it up to date every day.
"And it worked perfectly. All of that was done by Claude in the background, over weeks of background tasks, without me having to do anything besides send it another article on how to do two-bin systems for this type of thing. It has become unimaginable to use a system like an ERP without MCP access and all the AI features you've built."
— Sam Alter, Founder
"As the numbers get bigger, material costs get bigger, and production labor gets bigger, having all of it be as accurate as possible in a single system, without updating Finaloop outside of Fulfil, matters. I couldn't imagine doing that without hiring a full-time person to manage it. Now we have an agent that stays on top of it every day."
— Sam Alter, Founder
Fulfil's built-in AI agent can now act on its own recommendations instead of handing back a list of tasks.
"It can act on its recommendations, not just say 'here's what you should do, here's a list of thirty tasks to implement.' It'll just do it for you now."
— Sam Alter, Founder
One Shipper, 500% More Orders
This is Atlas's fourth Black Friday on Fulfil. The shipping team hasn't grown. The same one person handles all of it.
"We have the same shipping resources: one person who has managed a roughly 500% increase in shipments. All of that is done through automation, with automation rules and shipping batches."
— Sam Alter, Founder
Last year he shipped about 750 orders in a single day. This year the goal is around 800. Training hasn't changed and no new documentation has been needed.
"Frankly, the Pick Pack app in Fulfil is really good. I haven't used a lot of other pick-pack apps, but this one works really efficiently."
— Sam Alter, Founder
Atlas uses Fulfil's WMS, automation rules and batch picking to keep shipping volume off the founder's plate.
Software That Keeps Getting Better
Sam compares Fulfil to other tools he uses that have stopped improving.
"Everything we've ever complained about with Fulfil has been fixed, or at least addressed with a workaround if it wasn't fixed. That's very unique for a software company, to actually listen to its customers in a way that makes things better. Especially with ERP software, most companies are pretty resistant to change, and you aren't."
— Sam Alter, Founder
Not having to re-platform matters just as much.
"Not having to change our business processes every year as the business has grown has been massive. ERP implementations are painful and tedious. Never having to think about doing it again has helped us keep investing in Fulfil, without feeling stuck on a legacy platform that doesn't grow with us."
— Sam Alter, Founder
An ERP Costs Less Than an Employee
Sam is upfront that Fulfil isn't cheap. He compares it to a hire, not to other software.
"I think the most obvious thing that isn't obvious to everybody else is that the cost of Fulfil is high, for sure. I don't want to say it's not high. But when you compare the cost of Fulfil to the cost of an employee, Fulfil is going to be with you much longer than most employees. And the continued investment Fulfil makes means it's so much more powerful than an employee. Not having to continually train and invest in employees to fix broken software processes saves a lot of time and money over time."
— Sam Alter, Founder
"When I tell my business-owning peers that I use a very expensive ERP, most of them can't wrap their heads around the cost. But they also have five times more team members than I do, and I can't wrap my head around their management overhead. I'd rather invest in a leaner team and better software than a bigger team and worse software."
— Sam Alter, Founder
The Results
Using Fulfil, Atlas Pet Company has been able to:
- Run manufacturing, inventory, shipping and accounting with four full-time employees, plus the founder and one part-time teammate
- Ship 500% more orders with the same one person, using automation rules, shipping batches and Fulfil's Pick Pack app
- Replace a head of ops hire with Kai, an AI agent built on Fulfil's MCP and Claude Code, running for about six months
- Build a production planner in three weeks on Fulfil cycle data, after six months and $9,000 with a developer went nowhere
- Run a lifetime repair program from a custom portal built on Fulfil and Shopify
- Move accounting from Finaloop into Fulfil, kept up to date by an agent instead of a full-time bookkeeper
- Skip a $248K, nine-month NetSuite project and build on Fulfil without paying for service hours
- Make and ship from one system, instead of selling products to itself between Katana and ShipHero
See related: ERP for pet supplies brands
See related: How Grunt Style built 25 micro apps with Fulfil and Claude