12 Laid Off vs. 3 Added: How the 'First Task' Decision Shapes Everything

[[IMG:hero]]

Every time a story drops about a small business laying people off because of AI, the comments section loses its mind. Cold. Heartless. Zero responsibility. I got DMs asking what I thought.

Here’s my honest take. Before we argue about whether to cut or not cut, there’s a more fundamental question nobody’s asking: which task did you hand to AI first? Two companies in the same TIME feature ended up with completely opposite outcomes, and that one decision was the fork in the road.

One cut 18 people and freed up $250K a year. The other added the equivalent of 3 full-time employees in capacity — without firing anyone. Both called it a productivity win. The numbers go in the same direction. The actual experience inside those companies is opposite.

The choice readers need to make today isn’t “cut or don’t cut.” It’s: how do I pick the first task in my own business? That’s what we’re going to work through.

The Same TIME Feature. Two Completely Different Outcomes.

The anchor here is TIME’s May 14, 2026 piece, “The Small Businesses Already Replacing Workers With AI.” Two companies in that piece sit at opposite ends of the outcome spectrum.

Company one: Sonora. An online guitar school with a CEO running a 48-person org. After deploying AI agents, they laid off 12 “setters” — the outbound customer acquisition team — along with portions of sales management, customer onboarding, and operations. Headcount: 48 to 30. Revenue: held. Annual savings: roughly $250K. The implementation path was SaaS consolidation — replacing HubSpot, Calendly, Vimeo, and DocuSign with internally built custom tools, then centralizing customer data so AI agents could work cleanly across it.

Company two: Hospitable. A short-term rental property management platform, CEO Pierre-Camille Hamana. From December 2025 to May 2026, they increased AI spending by 50% — equivalent in headcount terms to 3 full-time employees. What is AI doing internally? 90% of code generation, 70% of customer support query handling, finance payment prioritization assist, and marketing campaign operations. Through all of that, Hamana said publicly: “We haven’t laid anyone off.” Instead, they simply slowed hiring.

Reading this as “reduce vs. expand” misses the point. Both companies genuinely increased output through AI. What’s different is whether they realized that gain by removing people or by amplifying the people who stayed. I call these the reduction model and the expansion model — and the choice between them is deliberate.

[[IMG:1]]

What I notice in TIME’s editorial judgment: they ran both companies in the same piece. The framing isn’t “AI steals jobs.” It’s closer to “the outcome is a choice the company makes.” The question to take away isn’t which side is correct — it’s whether your business is making that choice consciously.

The Real Fork Between Reduction and Expansion

“It just depends on your cash flow situation,” you might be thinking. Partially true. But not sufficient.

From what TIME described, I see three variables that drove the split — and they’re not about size.

Variable one: which tasks went to AI first. Sonora handed B2C touchpoint work to AI first — setter calls, onboarding coordination, outreach sequences. Labor-intensive, customer-facing, highly repetitive. When you hand that category to AI, headcount reduction is the natural downstream. Hospitable handed code generation to AI first. That’s a high-effort, back-office function that customers never see. Because AI absorbed the invisible load, there was no pressure on the visible team — sales, support, customer relationships stayed intact. The sequence changes everything.

Variable two: whether AI was framed as growth investment or cost reduction. Hamana positioned the AI spending increase as investment in business growth. That framing kept the math clean — even as costs went up, the return made the ratio work, and headcount slowdown replaced layoffs. At Sonora, the lead framing was “$250K in annual savings.” The moment you frame AI as cost reduction, payroll becomes the obvious target. The language you use in the announcement shapes what’s politically possible inside the company.

Variable three: what happened to the people who remained. At Sonora, the remaining 30 people shifted into AI oversight and operations roles. At Hospitable, existing employees weren’t reassigned — the org just grew slower instead of adding headcount. Same category of change but very different experience for the humans in the room.

Small business owners know the invisible cost of a departure: manuals that never got written, implicit knowledge that scatters, culture that thins. Hospitable didn’t pay that bill. That’s worth something.

The split between reduction and expansion came down to task type and internal framing — not budget, not headcount, not industry. Both are controllable. The moment you conclude “Sonora was bigger, doesn’t apply to me,” you’ve surrendered your right to choose.

[[IMG:2]]

I saw a version of this play out with a client. An online school owner with 5 staff introduced AI and announced it to the team as “cutting costs.” Three months later, her sharpest marketing person quit. “I figured I was next,” he said.

The owner was floored. “That’s not what I meant at all.” But she’d said “cutting” first, and that’s what landed. This hits harder at small scale — the founder’s exact vocabulary sets the emotional weather for the whole room. Frame AI as a cost story, and your best people start making exit plans.

The “Unused AI” Problem: What US Data Actually Shows

Step back from the individual cases for a moment. For all the coverage, AI adoption rates in the US are genuinely lower than the noise level suggests.

U.S. Census Bureau data cited in the TIME piece: only 18% of businesses report using AI for any business function. An Anthropic report from March 2026 noted that AI models currently handle only a small fraction of tasks they’re technically capable of performing.

Two opposite reads on that number. One: “Most people aren’t using it yet, no rush.” The other: “If that much runway is uncaptured, whoever moves first can own it.” I take the second read. Once AI cleanly handles a category of routine work at a company, the need to hire for that category disappears — permanently. The hiring market reshapes quietly underneath.

Big company moves confirm the direction. Snap announced in April 2026 that it would cut 1,000 employees — 16% of full-time staff — citing AI-driven efficiency. Layoff tracking services show an increase in AI-attributed job cuts starting in 2025, though methodologies differ enough that exact counts should be treated as estimates.

[[IMG:3]]

One thing to watch: don’t get absorbed by the layoff numbers themselves. Analysts are mostly reading this not as “AI replacing humans in real time” but as “companies building the capital base to fund the next phase of AI infrastructure.” The current numbers reflect a reallocation of capital, not a scoreboard for humans vs. machines.

For small business owners, solo operators, and side-hustlers, the signal is simpler: the live question isn’t “will AI take my job” — it’s “am I building the capital base to reinvest in the next phase?” That’s the operating logic behind both Sonora and Hospitable, just applied differently.

Anthropic released Claude for Small Business in May 2026 — 15 domain-specific AI agents that give small operators a decision map for where to start. Nagi’s breakdown is here. “The tools exist, now where do I start?” has officially become the real question.

Three Questions to Answer Before Picking Your First AI Task

From here, it’s about your business. Three questions to work through before you choose the first task to hand off.

Question 1: Which tasks in your business run on the same steps every week? This is the starting inventory. A task that takes 10 minutes a day, 5 days a week is about 3.5 hours a month. Stack 20 of those and you’ve got 840 hours a year — roughly 0.4 FTEs. Small business owners who feel like strategy never gets time are usually burning it here, across dozens of tasks that feel small individually.

Question 2: Of those repetitive tasks, which ones have low proprietary value? Low proprietary value means: anyone could do this and the output would be nearly identical. Invoice generation, contact list cleanup, email template replies, meeting note formatting, social scheduling, metric transcription. Contrast that with: client relationship building, pricing negotiation, brand direction, new hypothesis generation. First category goes to AI. Second stays in your hands. Sonora handed the first category and contracted. Hospitable handed the first category and absorbed the gain through slower hiring.

Question 3: When AI frees up that time, what are you ready to do with it? This is the most important question — and the one most people skip. Operators who don’t have an answer tend to redirect freed time toward cost cutting. Lock in your answer first: “new customer outreach,” “product development,” “deeper existing client relationships.” Whatever it is, define it before you start the AI conversation. With that anchor in place, the reduction framing loses its gravity.

Work these three questions in order and you’ll naturally land in expansion model territory. Sequence matters: answer Question 3 first, then return to Question 1. If “I’ll use that time for new customer conversations” is the answer, research-intensive work goes to AI first. If “product development” is the answer, operational routines go first.

For solo operators, this filter is the difference between looking back in six months and seeing an expanded business vs. a cheaper one. I’ve written about the inflection point where solo workflows need to change in The Three-Axis Check Before You Hit Solo-Founder Capacity. Cost sequencing with ROI priority is in AI Stack for Solo Founders: What Does a Month Actually Cost?. Pair those with today’s framework and the map is complete.

How to Choose Your First Task: A 3-Step Framework

Everything above, compressed into three executable steps you can run this week.

[[IMG:4]]

Step 1: Inventory. Write it down — by hand, in Excel, doesn’t matter. List every task you repeat every week. Include anything under 10 minutes. Include the quick checks you do three times a day. Expect 20–40 tasks. If that feels like too many, that’s the point. The reason you feel busy is about to become visible.

Granularity standard: one step-by-step process that would fit on a single page as written instructions for someone else. “Email management” is too coarse. Break it into: “first-response to new inquiries,” “auto-confirm for registrations,” “first-touch for complaints.” Stay coarse, and handing it to AI produces confusion, not results.

Step 2: Score. Three axes, 10 points each. ① Time score: 1 point for under 1 hour/month, 10 points for 10+ hours/month. ② Frequency score: 1 point for once a month, 10 points for daily. ③ Proprietary value: 10 points if the output depends specifically on your judgment or relationships, 1 point if anyone could produce the same result. Add the three scores — don’t multiply, add. Max is 30.

Key rule: a task scoring 10/10/10 does not go to AI, even though it dominates your schedule. That’s the core of your business. The right candidate: proprietary value 1–2, time 6–8, frequency 6–8. High cost to you, low replacement risk. That’s the target profile.

Step 3: Run a one-month experiment on one task. Pick exactly one task. Not two. One. Give it to AI for 30 days. Track three questions: Did it take less than half the time? Did quality hold? Did I actually use the freed time the way I said I would? All three yes → move to the next task. Any no → either bring it back or change how you’re handing it off.

The virtue of this approach: failure is small. One task, one month — if it doesn’t work, you restore it and move on. Sonora’s 18-person restructuring in a single wave isn’t the model for small operators. Small scale is the advantage here: test and stack, rather than commit and restructure.

My own first task to hand off was meeting note formatting for clients. About 10 hours a month, near-zero proprietary value. After one month, I had 10 extra hours. I put those into new client proposal prep. Three months later, one new contract closed. Pick the right first task and the chain reaction runs forward. Pick the wrong one and you’re back to doing it yourself inside two weeks.

Wrapping Up

“Should AI cut people or not” is the wrong starting question. Three things need to be decided before you get there: how to choose which task to hand off first, how to frame that choice to your team, and where to direct the time you recover.

Small business owners, solo founders, and side-hustlers share a common blind spot: the task inventory has never been done. The more you’re doing alone, the more acute this is. Before any AI conversation, visualize your own week. Write out 20–40 tasks. Half your problems become clear at that step — because the shape of “I’m too busy” becomes legible for the first time.

The reduction model isn’t wrong. Sonora’s decision follows a coherent logic, and there are business shapes where contracting makes sense. But at small scale, the expansion model almost always gives you more options. You don’t need to grow headcount to grow capacity. That’s the real gift AI brought to small business in 2026.

Not “someday.” This weekend, two hours, start with Step 1. Nothing changes if you don’t move. Move, and the first task shows itself within a month.

References

• The Small Businesses Already Replacing Workers With AI (TIME, May 14, 2026) — primary source
• Nagi’s breakdown of Anthropic Claude for Small Business — the 15-agent structure
• The Three-Axis Check Before You Hit Solo-Founder Capacity — decision/outsource/systematize map
• AI Stack for Solo Founders: What Does a Month Actually Cost? — ROI-ordered cost design