If you run a $5–50M company and you're still answering employee texts at 10 p.m., this is for you.
Picture your week in 90 days: the inbox handled before you sit down, the report written before your L10, the coordinator you were about to hire never posted. That's the goal. The mechanism is a done-for-you AI install, built by a CEO who did it inside his own non-tech business first, and has never sold it outside his company until now.
You've tried ChatGPT. Maybe you bought a tool. Nothing stuck, because the hard part was never the technology. The hard part is knowing what to build first, building it fast enough to matter, and getting your people to actually use it. That's a method, and nobody's handed you one.
You're not behind. You're early. Here's the whole road, and where the install takes you.
You type a prompt, it answers. Useful, but it's still you doing every task, one window at a time.
One human, a team of agents: a chief of staff, a CFO, a recruiter, an office admin, each with its own computer, working while you sleep.
Your whole team working alongside agents inside Slack or Teams, each agent scoped to what that channel is allowed to see.
The agent finds the missed quote, the duplicate ticket, the unhappy client, and pulls the right person in: "Human, sign off on this."
The install gets you to stage two in week one and has your team working at stage three by day 90. Stage four is what your internal owner builds after.
Divide what you take out of the business in a year by a 2,080-hour working year. Most CEOs guess a few hundred dollars. It's usually more. Then divide that by four: anything that can be done for less than that rate shouldn't be on your desk. And because Johnny finds 1% of top line in phase one of every install, your revenue tells you what's waiting.
That's not overhead. That's your time and your margin, and they're the first two lines on your ledger.
Nothing you enter here leaves your browser.
Give me your email and I'll send you the exact one-page sheet my team uses to decide which subscriptions to kill, what replaces them, and in what order. It's the first thing we do in every install. Most companies find a five-figure line on the first pass.
Not a roadmap. Not a workshop. Working systems, on your own instance, owned by your own people.
An agent that runs your inbox, chases follow-ups, tracks commitments, and drafts the weekly report before you've had coffee. Live in your inbox in the first week.
Built around your scorecard or your worst recurring headache: proposals, quotes, client status, onboarding, invoicing. Whatever's eating your team's time.
Every subscription you're paying for that the new stack replaces, with the replacement in place before we cancel anything. Software is no longer a moat. What you do with it is.
A dedicated instance of the EOS-native platform Johnny runs his company on: scorecard, L10s, rocks, 1-on-1s, client pulse. It forces the cadence at the department level, across offices and countries, so what the coach taught actually gets run. Twelve months included.
Azure AD or Google SSO. Teams or Slack. 365 or Workspace. Service accounts your IT manager controls. Private agents for you, restricted agents for the team, so nobody asks the CFO bot for the annual revenue.
One named person who runs the stack after day 90, plus a champion in each department who shaped their own workflow and will keep pulling it forward. You don't become dependent on Johnny, and you don't depend on one person either.
The bi-weekly tips-and-tricks meeting that keeps the whole company compounding. Installed and run twice with Johnny in the room.
Three things that happened at Johnny's engineering firm in the last few months. Same agents, same method, first try.
A partner who owns big quotes was missing them in his inbox, even with a full-time assistant dedicated to chasing him. An agent now reads his email against the ticketing system every day and posts a red, yellow, green tracker in Slack with a screenshot of each thread. It caught a quote that had sat unanswered for a month.
The COO needed licensed engineers who fit ten criteria: location, years, industry, no job-hopping. A recruiting agent reads LinkedIn every day, builds the list, and learns from every "yes" and "skip" she replies with, so tomorrow's list is better than today's.
"I'm hosting the group. Look up who's coming, check their food preferences, suggest something." Four agents coordinated it: one researched the guests, one picked the vendors and opened the accounts, one handled the messages, one paid and sent the receipts to AP. It arrived on time.
Numbers are as reported by the people using the agents, not an audit. That's what your ledger is for.
The classic consulting shape is four weeks of discovery, a big decision gate, a build phase, then a handoff. That works for an ERP rollout. With AI it's the wrong shape: by the time it ships, the tools have moved on and nobody finds it useful.
We put a minimal version in your people's hands in the first week, let them use it in their actual day, see where it saves them time, take their feedback, and iterate. Each round makes the tool more useful, people get more done, and because they shaped it, they own it. Then we build the next layer on that foundation. You don't boil the ocean. You compound.
AI isn't a siloed implementation you hand to one maintainer. The real unlock is people across the company adapting it to their own work. You need champions in each area, not just one person who knows how to redeploy.
How it's introduced decides whether you get that. A big top-down process reads to the team as "the boss is bringing in AI to replace us." People go quiet, you get polite feedback instead of real feedback, and adoption stalls. When they're part of building it from the start and see it making their own job easier, they pull it in rather than resist it. The tools are the easy part. Adoption is where the value shows up or doesn't, and it can't be a change-management step at the end.
Itemized the way an engineer would price a job. Every line is something you could buy separately. You're not buying them separately.
50% at signing, 50% at day 45. Never hourly.
Sign and pay before December 31, lock $85,000, and start late January. Anyone who waits pays $120,000.
Some CEOs prepay in the current calendar year. Ask your CPA whether that helps you.
Two CEO sessions and a ranked map of where your time and money leak. If someone else already built you agents, this is the review. If you go on to the install, the whole fee comes off the price.
We agree on $180,000 of value, line by line, before we start. If at day 90 the ledger only says $150,000, you still pay the second half, and I keep working for free until the ledger says $180,000.
One condition: show up to the four sessions and give us the access we ask for in week one. The rest is on me.
You and I might disagree on a number. We'll agree on the method. Every line is priced the way you'd price a hire: if a US staffer did this job for this many hours a year, it costs this much. Subscriptions count at what you pay. Hours you get back count at what your hour is worth. You sign off on each line before it goes on the ledger, and nothing counts until it's running.
| Line | Priced as | Value |
|---|---|---|
| Quote trackerReplaces a dedicated assistant chasing one partner's inbox | Half an admin role, loaded | $40,000 |
| Three subscriptions cancelledReplaced by the platform and one built workflow | Annual invoices | $36,000 |
| Recruiting shortlist agentDaily LinkedIn screen against ten criteria | VA hours, loaded | $24,000 |
| Ten hours a week of your inboxHandled by the chief of staff | Your hourly value × 500 hours | $120,000 |
| Ledger at day 45 | $220,000 | |
Your AI chief of staff is live in your inbox. You feel it before your next L10.
First subscription cancelled. Operator Audit complete. Ledger v1 on the table, ranked by payback.
Both workflows in use. Ledger checkpoint: every line signed off. Second payment due.
Platform live, owner and champions trained, 1% Lab running without Johnny. Ledger at $180k, or we keep going.
You don't need someone who read about AI. You need someone who had to make it work in a business like yours, with a team that didn't ask for it and payroll due on Friday.
That's Johnny Vu's day job. He's CEO and co-founder of Lam+Tea Engineering, a 100+ person MEP engineering firm with offices in Los Angeles, Las Vegas, and India. Before that he was a digital transformation officer at a Fortune 100 company and ran supply chain and transformation work at Walmart, CVS, Sony, Williams-Sonoma, Sam's Club, and JCPenney, back when "AI" was called machine learning and automation.
Everything in this install exists because he built it for his own company first: an AI chief of staff that runs his inbox, an office admin agent, a custom CRM that replaced Salesforce, an HR app, an EOS platform that replaced the one he was paying for, and a bi-weekly lab that taught a team of engineers to use all of it.
He's not selling AI. He's handing you the method that worked in a business that looks a lot like yours.
Most AI consulting sold to companies your size is one of the first two columns. Ask whoever you're talking to which one they are.
| Typical AI consultant | Training and workshops | The Lean Operator Install | |
|---|---|---|---|
| What you get | Scoping interviews, some agents built, a training session on how to modify them | Slides, a prompt library, a lunch-and-learn | Working systems on your own instance: chief of staff, workflows, kill list, platform, trained people |
| First result | After the discovery phase, usually week five or later | Whenever your team gets around to it | Day 7. A chief of staff live in your inbox |
| Who builds | They build, then hand off | You do | Johnny's team builds, with your people in the room so they own it |
| Who runs it after | A champion you have to find and fund | Nobody, usually | An owner we train, plus a champion per department |
| Guarantee | None | None | $180,000 in agreed value or he keeps working for free |
| Who's done it before | Worked in tech, advised companies | Reads about it | Runs a 100-person non-tech company on it every day |
| Pricing | Hourly or phased, open-ended | Per seat or per day | Fixed. $85,000. Never hourly |
Compared to what? One admin hire is $80,000 a year, every year. A fractional exec is $10–25k a month and ships nothing. An EOS coach is $8,000 a session, four times a year. This is one-time, done in 90 days, and guaranteed against a ledger you sign.
You still pay, and Johnny keeps working for free until the ledger says $180,000 in value you've agreed to. That's the deal. In practice the first bloated software stack and the hours your inbox takes usually get there before day 45.
No. Hourly pays Johnny to go slow. A fixed price pays him to finish.
Two options. Grab the kill list and run it yourself. Or buy the Operator Audit on its own for $15,000: two sessions, a ranked map, and the full fee credited if you continue. If someone else already built you agents, the audit is the second opinion.
Ask them three things: what's live by day 7, who owns it after they leave, and what the guarantee is. If the answers are "after discovery," "a champion you'll need to find," and "none," you're in the first column of the table above.
Neither is Johnny's. That's the point. His team does the building; your team's job is to show up four times, name one owner, and let a champion in each department shape their own workflow.
Fine. Azure AD for sign-on, Teams for the agents, 365 for mail and calendar. Your IT manager creates the service accounts and controls what each agent can see.
Because Johnny is still running his own company. Three is what his build team can carry alongside the day job without either side getting a worse product. The founding price is lower because you'll be the proof: a video testimonial once the value lands.
Yes. Sign and pay before the cohort closes and you lock $85,000 with a late January start. After the cohort, the price is $120,000 without the bonuses.
Your owner and champions run it. If you want Johnny to stay involved, there's a light monthly retainer, capped at three clients. Your platform instance is included for the first year and licensed after that.
He'll come with four questions about where your time and money are going, and he'll tell you what he sees, whether or not you turn out to be one of the three. No deck, no pitch.