The lean-team era: forty people, a quarter-billion.

Dude Wipes does roughly a quarter-billion in sales with about forty people. We run Canvas & Ivy’s entire customer-service operation with software we built ourselves and almost no humans in the queue. These are the same story told twice: the constraint that used to define a consumer brand — how many people you could afford to hire — has quietly stopped binding. This is a piece about what running lean actually looks like in 2026, why Dude Wipes is the canonical example, and how we replaced a customer-service team and its entire SaaS stack with an in-house AI system. Plus the honest part: where lean breaks.

The canonical lean brand

Dude Wipes is the cleanest proof that lean is a competitive weapon, not a cash constraint. Four childhood friends, one Shark Tank check from Mark Cuban in 2015 (the only outside money they’ve ever raised), profitable since 2016, and now the number-one flushable-wipe brand in America — having passed Cottonelle, a Kimberly-Clark brand whose parent does ~$20B in sales and is headquartered, literally, across the street from Dude’s Chicago office. They did it with a team of about forty against a category dominated by P&G and K-C. Riley’s framing is the whole playbook: “That’s how you compete with a Goliath — being completely differentiated… they can’t compete with our spirit and our creativity, how fast we move and the things we say.”

What changed: AI made lean repeatable

Being lean used to require a genuinely exceptional team — a Moiz Ali at Native, a Sean Riley at Dude. AI is turning the exceptional structure into an installable one. Dude Wipes now gives staff Claude accounts and regular AI training, and runs an internal ops tool (“Diesel Dan”) that saves 10–15 hours a week on logistics work — while, tellingly, the CMO says they’re adding marketing headcount after years of deliberate leanness. That nuance matters: AI didn’t eliminate the team, it changed which roles are worth a human. Shopify codified the same logic in Lutke’s 2025 memo: “Before asking for more headcount and resources, teams must demonstrate why they cannot get what they want done using AI.” That sentence is now the default operating posture for a generation of lean brands — and it’s exactly how we run.

How we replaced our CS team with in-house AI

Here’s the part that’s genuinely advanced, and the part founders ask us about most. At Canvas & Ivy we did not buy a help-desk SaaS, bolt an “AI add-on” onto it, and staff a team to babysit the queue. We built the customer-service system ourselves — an AI agent wired directly into our order, fulfillment, and returns data — and retired both the human queue and the stack of subscription tools it used to run on. The agent drafts and resolves the long tail of “where’s my order,” “can I change this,” “how do I…” with a human reviewing only the genuine edge cases. The results tracked what the best operators report: resolution time collapsed, and — the counterintuitive part — customer satisfaction went up, because instant, accurate, 24/7 answers beat a same-day email from a tired human. This is the third discipline of the Lampwork Operating Standard in practice: encode the repeatable work as software, keep the team senior and small, and spend humans only where judgment is required.

The broader market validates the direction. Intercom’s Fin resolves ~67% of tickets autonomously; AI-CS startups like Sierra (~$15B valuation) and Decagon (~$4.5B) are pricing on outcomes — you pay only when the issue is truly resolved. The difference in our model is ownership: by building it in-house against our own data, we don’t pay per-resolution rents to a vendor, and the system gets smarter on our brand specifically. That’s the LSOP thesis — a procedure, encoded once, that ships across every brand we own.

We didn’t buy AI customer service. We built it, pointed it at our own data, and deleted the queue and the SaaS bill at the same time. The CSAT went up. That still surprises people.

The honest caveat: where lean breaks

Lean is a discipline, not a headcount-zero fantasy — and AI customer service done carelessly is a trust-destroyer. The cautionary tale is Klarna: it cut ~40% of its workforce on AI confidence, declared the assistant did the work of 700 agents, then publicly reversed — “we went too far” — and began rehiring humans after service quality and trust eroded. The lesson isn’t “don’t automate.” It’s that automation has to raise the customer experience to earn the headcount it replaces; if it lowers it, you’ve traded a cost line for a churn line. The genuinely lean consumer brands — Dude Wipes adding marketers, us keeping a human on the hard CS tickets — all keep people exactly where judgment, taste, and relationships live. Everything else is software.

Sources: Inc. (Dude Wipes $1B); Forbes (Dude Wipes); Digiday (Dude Wipes AI); CNBC (Lutke memo); Intercom Fin; Bloomberg (Klarna reversal); TechCrunch (Sierra, Decagon valuations).