Six ways to grow without adding headcount.
Each one is something a big brand staffed a department for. Each one starts with a tool small enough to ship in weeks, tested on your own data, and grows from there. Below: what it is, what it used to take, what the first build looks like, where it’s running today, and what I’d build next.
1 of 6
A company that remembers.
Every policy, dealer term, season review and customer conversation in one place your team and your agencies can ask. Knowledge stops leaving when people do.
Your policies, dealer terms, season reviews, and the customer conversations that taught you something, gathered in one place. Anyone on your team, or at your agencies, asks it a question in plain words and gets the answer with the source it came from. When someone leaves, what they knew stays.
Used to take: a project management team and an intranet nobody read.
The first build
One body of knowledge your team asks about every week, made answerable. Dealer terms and policies are a common start.
- Example
Team Answer Desk. Built from your real SOPs, policies, and message threads, not another wiki to keep up to date. It reads what you already have in Notion, Confluence, Google Docs, and Slack. When you ask how you handle a dealer return in Canada, it answers the way your best person would, with the source cited. - Example
Voice-of-Customer Digest. Reads across your helpdesk, reviews, and returns comments at once. It clusters the fragments into the handful of things customers are actually telling you, and routes each theme to the person who can fix it.
Where it’s running
Loop Trails runs on a knowledge base of 850+ articles that check themselves. No client build of this one yet.
2 of 6
Every customer and dealer treated like your first.
Each inquiry is heard, sorted, and assigned to the right person within minutes. Your team still writes every reply.
Every email and dealer request that lands in a shared inbox is read, sorted by what it is and how urgent it is, and put in front of the person who owns it, in the tool they already work in. Nothing sits for a week because it looked like someone else’s job. It sorts and assigns. It doesn’t answer.
Used to take: a large CS team and a host of rep agencies.
The first build
Your shared inbox, read on a schedule, with each message sorted and assigned. It runs read-only beside your team until the results are checked, and only then is it allowed to write anything.
- Example
Warranty Claim Intake & Routing. A customer emails photos of a cracked weld. Working alongside your helpdesk, it reads the claim, checks the product and the purchase, drafts the disposition (replace, repair, decline), and routes it to the right person to approve. - Example
Wholesale Order Desk. A retailer sends a purchase order as a PDF, a spreadsheet, or a plain-text email. The Order Desk matches it to your catalog, flags exceptions, and writes clean order lines into your ERP, with a person approving before anything posts.
Where it’s running
Super Pacific builds truck campers. Their shared inbox is read every five minutes. Since September 21, each sorted message becomes a ticket in their ERP. It sends zero emails to customers. Before it was allowed to write, ten live calls in a row were hand-checked with none wrong, and it scored 51 of 53 on a blind test of real mail it had never seen. Two models have to agree before it acts. Triage time per email is down 92%.
3 of 6
A year of creative without a bigger team.
One line of intent becomes the brief, the assets, the approvals and the results, tied together from start to finish.
Someone writes one line: what you want to make, and why. That line becomes a brief, a folder for the assets, and a designer assigned to it. Feedback and approvals happen in Slack, where your team already talks, and every piece stays tied to the brief it came from.
Used to take: a creative team and/or agency, plus project and traffic managers.
The first build
The intake: one line in; a brief, an asset folder and an assignment out, with approvals in the chat tool you already use.
Where it’s running
Süti makes iPhone accessories. One row in a Google Sheet becomes a brief, an asset folder and a designer assignment, with feedback and approvals in Slack. Administering creative now takes 50% less time.
I hold equity in Süti.
Next up
What I’d build next under this one. None of these is running at a client yet.
- Your customers already wrote your ads. Your reviews, support tickets and comments hold the exact line that answers a buyer’s real complaint. This reads them, finds the customer who already said it, and puts that quote in the brief. The ad carries proof instead of a claim.
- Listen where your buyers complain. People say what’s wrong with the gear they own on Reddit and in forums, and what they were doing when it failed. This watches those threads for your category and turns the complaints worth answering into briefs your team can pick up.
- Enough ads to find the winner. Before you have a winner, the job is coverage, not polish. One brief becomes dozens of concepts cheaply, so you learn which angle works before your designers spend a week on it. Then they go deep on that one.
4 of 6
The whole business, every morning.
Know on Monday what you used to learn at quarter close, in a database you own. Decide in a day, not a season.
Sales, ad results, and stock, pulled from the tools you already pay for into one database each morning, so Monday’s numbers are already waiting. You stop waiting for the quarter to close to find out what happened.
Used to take: a data team, analysts, and a data warehouse.
The first build
One source, every morning: usually your ad results, landing in a database that’s yours, not an agency dashboard.
- Example
Wholesale Sell-Through Digest. Every retailer’s sell-through report gathered into one ranked call list: who’s selling, who’s stalling, who your rep should call this week. - Example
Monday brief. The week’s numbers across your tools, plus the one next action each number implies. The paid-media version says what last week’s spend earned, as a decision, not a chart.
Where it’s running
Daily Meta results land in a database Süti owns, not an agency dashboard. Next: closing the loop between ad performance and creative, then automating the new creative brief.
I hold equity in Süti.
Next up
What I’d build next under this one. None of these is running at a client yet.
- Don’t sell what you can’t ship. When an ad takes off, this checks what’s in stock and what’s on order before the budget goes up. If you can’t fill the orders, it eases the ad down rather than running it into a stockout, because pausing a winner costs more than slowing it. A person on your side confirms the change.
- A planning partner to your retailers. What’s selling on your site and what’s working in your ads, turned into something you can hand your buyer at the retailer. You show up with a point of view on their next order, not just a line sheet.
5 of 6
The brand the AI assistants recommend.
When a buyer asks ChatGPT or Perplexity what to buy, are you the answer? I measure it, then fix what keeps you out. Most brands have never checked.
I ask the AI assistants the questions your buyers ask, the way they ask them, and record who gets recommended. Then I find what keeps you out of the answer, fix what can be fixed, and measure again. I can’t promise you become the answer. I can show you where you stand and what moves it.
Used to take: nothing. This one is new for everyone.
The first build
A baseline: the questions buyers in your category ask, run across the major assistants, showing where you appear and where you don’t.
Where it’s running
Not yet on a client’s. On a run on September 17, 2026, four AI assistants named Loop Trails 15 times out of 16 when asked by name, and 0 times out of 64 when asked the way a buyer asks. On the same run, fewer than 1 in 10 of the brands the assistants recommended also ranked on Google’s first page. Ranking on Google and being recommended by an assistant are different problems.
Next up
What I’d build next under this one. None of these is running at a client yet.
- What moves the number. Three things move it: pages built so an assistant can read them, your stories picked up by other sites, and other people mentioning you. The baseline shows which of the three you’re missing, and that’s the first fix.
6 of 6
A team of builders.
Your people make the tools they need, inside clear rules. Nobody waits on IT, and nobody has to become the AI person.
The people closest to the work build the small tools they need, inside rules you set together: what data they can touch, what has to be reviewed, and what can never reach a customer without a person saying yes. I set up the rules and the foundation.
Used to take: an IT department, program managers, executive sponsors, Op Ex and Cap Ex budgets and months, if not years.
The first build
The rules and a starting kit for one team, plus the first tool they build with it.
Where it’s running
My own shop is the working model: 27 apps that run everything from the daily agenda to coding.
Next up
What I’d build next under this one. None of these is running at a client yet.
- One seat, not six. When requests for AI tools go through one place, the company pays for one license instead of one for everyone who uses a tool twice a month. That same place is where the rules live: what each tool may read, and what has to be reviewed before it goes out.
Hospitality
Same machine, different nouns.
This page is written in retail nouns. Swap in bookings, room types, and occupancy and the six still apply. If you run a lodge, a resort, or a restaurant group, read it with your nouns.
Pick the one you’d do first.
Bring that to a forty-five-minute working session. You’ll leave with a first build named, and a fixed-price quote follows in writing.