TL;DR: Hrag Kalebjian, owner of Henry's House of Coffee in San Francisco, used Google Gemini for three tasks: building a custom sales forecaster in Google Sheets, converting a dense instruction PDF into a print-ready customer guide in 20 minutes (a 95% time cut), and drafting his weekly newsletter from industry news summaries (Google, July 8, 2026). None of this required a marketing degree or a dedicated hire. It required knowing exactly what his business is and using AI to protect that position instead of chasing every new tactic. That's the FOCUS Strategy: find your position, then let the system defend it.
I spent years on a submarine watching junior officers try to do six jobs at once because we didn't have the headcount to do one job each. The ones who thrived didn't work harder. They figured out which one job actually mattered and let procedure handle the rest. Hrag Kalebjian runs a coffee shop, not a boat, but he's solving the same problem every $500K to $5M owner-operator faces: too many hats, not enough hours.
The Case Study, Straight
Henry's House of Coffee is a San Francisco institution that Hrag took over from his father. He built a nationwide subscription business alongside the physical shop, which means he's simultaneously a roaster, an accountant, and a marketer. Google profiled how he uses Gemini to claw back hours across three specific tasks (Google Blog, July 8, 2026).
One. Automated sales forecasting. Hrag needed a way to track daily revenue against his monthly average without buying expensive analytics software. He described what he wanted in plain language, and Gemini wrote the custom Google Sheets code for him, including a function to flag any day sales fell below average. He troubleshot errors by sending Gemini screenshots. Total setup time: about an hour. He now gets the report every morning without touching a formula.
Two. Instruction card redesign. The shop sells Armenian coffee kits with a dense, 15-step instruction PDF that nobody wanted to read at checkout. Hrag uploaded the old document to Gemini and asked for a clean, print-ready customer guide. "After maybe 20 minutes of me uploading and talking to it, I had a card that was ready to print," he said. The shop now creates customer-facing assets 95% faster than the old design workflow.
Three. Newsletter content from industry news. Hrag writes a weekly newsletter called "The Sunday Pour." Instead of drafting from a blank page, he feeds Gemini coffee industry updates and has it produce a first draft in his brand voice. Google notes that Pro and Ultra subscribers can schedule recurring actions so Gemini compiles weekly industry reports automatically, without Hrag having to go looking for news himself.
Why the Small Case Study Beats the Big One
Every AI vendor wants to show you a Fortune 500 logo. That's not useless, but it's not usable. You don't have a Fortune 500 data team. You have you, maybe two employees, and a task list that doesn't end at 5pm.
The data backs up why this matters. Fiverr's 2025 survey of nearly 6,000 small business owners found 70% spend less than five hours a week on marketing, even though they rank it as critical to growth (Fiverr, 2025). Enji's State of Small Business Report found 57% of owners with ten or fewer employees spend just one to five hours a week on marketing, squeezed between operations and everything else (Enji, 2025). Constant Contact's global survey found 42% of small businesses have less than one hour a day for marketing at all (Constant Contact, 2025).
Hrag isn't unusual because he found time. He's unusual because he stopped trying to find more time and instead removed the tasks that were eating it. That's the entire operator's dilemma. You don't win by adding hours. You win by cutting the work that never should have taken hours in the first place.
Goldman Sachs' 10,000 Small Businesses research backs this up at scale: 76% of small business owners surveyed are now using AI, 93% report a positive impact, and 84% cite increased efficiency as the primary benefit (Goldman Sachs, 2026). The U.S. Chamber of Commerce Foundation's Main Street AI Monitor found the dominant use case isn't automating jobs, it's augmenting the people already doing them, with 90% of task-level AI adoption showing up in writing and editing communications (U.S. Chamber of Commerce Foundation, 2026). Hrag's newsletter workflow is that finding in one shop.
The Owner-Operator's Real Constraint Isn't Money
Read Hrag's story closely and notice what's absent. He didn't hire a marketing agency. He didn't buy a $500-a-month design subscription. He didn't bring on a part-time social media manager. His constraint was never capital. It was hours, and more specifically, the hours he had left after roasting coffee, managing inventory, and running a subscription operation that ships nationwide.
This is the constraint most owner-operators misdiagnose. They think their marketing problem is a budget problem, so they either underinvest and watch their brand go stale, or they overspend on an agency retainer that produces generic content nobody on their team actually reviews. QuickBooks' 2026 AI Impact Report, built from survey responses of more than 34,000 business owners and anonymized data from 5.3 million QuickBooks businesses, frames the same finding at scale: the businesses gaining the most ground on AI aren't the ones spending the most, they're the ones directing existing hours toward higher-value work instead of busywork (QuickBooks, 2026). JPMorgan Chase Institute's transaction-based research on small business AI spending found monthly AI costs have actually fallen, from roughly $50 a month in 2019 to $20-30 a month in 2025, even as usage has broadened (JPMorgan Chase Institute, 2026). The tool got cheaper. The hours it buys back got more valuable. That's the trade every operator under $5M revenue should be making right now.
The FOCUS Strategy
I built the FOCUS Strategy for owner-operators who feel like they're doing twelve jobs badly instead of one job well. It's five steps.
F. Find your position. Hrag's position isn't "coffee shop." It's "San Francisco legacy roaster with a nationwide subscription business." That specific combination is the moat. Nobody else has his father's history and his digital distribution at the same time.
O. Outline the tasks that protect that position. Forecasting revenue protects the operations side. Clean customer materials protect the brand experience. A consistent newsletter protects customer retention on the subscription side. Every task Hrag automated maps directly to defending his actual position, not chasing a trend.
C. Cut everything that doesn't. He didn't ask Gemini to write ad copy for a paid campaign he doesn't run, or generate a social strategy that has nothing to do with his customer base. Discipline is as much about what you don't automate as what you do.
U. Use the tool as staff, not as magic. Hrag treated Gemini like a junior analyst: give it context, correct its errors, iterate. He shared screenshots when the sales-forecaster code broke. That's how you manage headcount you can't afford to hire. You still supervise the work.
S. Scale the win. Once the newsletter workflow worked, he didn't stop there. Gemini's scheduled actions for Pro and Ultra subscribers now compile his industry reports automatically every week, turning a one-time win into a standing process.
Three Prompts You Can Run This Week
Steal these directly from the case study and adapt them to your business.
- Forecasting: "Write a Google Sheets function that flags any day where my sales fall below my trailing 30-day average, and walk me through how to install it."
- Asset conversion: "Turn this internal guide, SOP, or instruction sheet into a clean, friendly, print-ready document for customers. Keep it to one page."
- Content drafting: "Read these industry articles or links and give me a three-bullet summary in a casual tone for a customer newsletter, matching this voice." Then paste a sample of your past writing.
None of these require code. None require a subscription beyond what you likely already have. They require twenty minutes and the willingness to treat the AI like staff you're training, not a vending machine you expect to get right on the first try.
What Happens After the First Win
Hrag's story doesn't end at the instruction card or the forecaster. The pattern that matters is what he did after each win: he didn't stop at one automated task and call the job done. He looked for the next task that was eating hours without protecting his position, and he automated that too. That's a discipline, not a lucky break.
Most owner-operators try one AI tool once, get a mediocre result on the first prompt, and conclude the technology isn't ready for a business their size. Hrag's account shows the opposite pattern: the sales forecaster took troubleshooting and iteration before it worked cleanly. The instruction card worked faster because he'd already learned how to give Gemini useful context. Each win made the next one cheaper to produce. That compounding effect, not the individual 95% time-savings number, is the real takeaway for owner-operators reading this from a business that isn't a coffee shop.
Doctrine Connection
This is Rule Two of the Founder's Doctrine: defend your position before you expand it. Most owner-operators try to grow their way out of operational drag. Wrong order. Fix the drag first with tools that protect the specific thing that makes your business yours, then expand. Hrag didn't use Gemini to become a different kind of business. He used it to become a faster, cleaner version of the exact business he already was. That's the difference between a tool that helps you exit on your terms and a tool that just adds another subscription to the pile.
FAQ
Q: Do I need Gemini specifically, or does this work with other AI tools? A: The specific tasks in this case study, custom code generation, document conversion, and content summarization from source material, are available across major AI assistants. Gemini's advantage in this case was native integration with Google Sheets and Docs, which most small operators already use. Pick the tool that lives inside your existing workflow, not the one with the flashiest demo.
Q: How much technical skill do I need to build something like the sales forecaster? A: None, based on Hrag's account. He described the outcome he wanted in plain language and used screenshots to debug errors. That's the whole skill required: describe, review, correct.
Q: Isn't 95% faster an exaggerated marketing claim? A: It's Google's own case study framing, and it applies to one specific task, converting an existing document into a print-ready asset, not to all creative work. Expect strong, not universal, time savings. A 15-step PDF becoming a print-ready card in 20 minutes is a real and repeatable result for similar document-conversion tasks.
Q: What's the risk of relying on AI-drafted customer materials? A: Voice drift and factual errors if you don't review the output. Treat every draft as a first pass from a junior staffer, not a finished product. Hrag still reviewed and approved the card before it went to print.
Q: My business isn't retail or e-commerce. Does the FOCUS Strategy still apply? A: Yes. The framework is position-first, not industry-first. Any owner-operator can name their specific market position, map the tasks that defend it, and cut the rest. The tools change by industry. The discipline doesn't.
*Disclosure: This article discusses a case study published by Google and independent third-party research for informational purposes. DEMG has no financial relationship with Google, Gemini, or Henry's House of Coffee. This is not a paid endorsement.*