What a $54K/year office admin actually does at a small yard — and what AI replaces

The pitch every AI vendor makes to a yard owner is some version of "replace your office admin for a fraction of the cost." It's a useful frame for a sales deck. It's a misleading frame for actually running a yard, because it implies a one-for-one swap that doesn't exist.

An office admin is not one job. It's about a dozen jobs braided together. Some of them AI does better than a human. Some of them AI does worse. Some of them AI doesn't do at all, and pretending otherwise is how you end up with a customer screaming at the loader operator on a Tuesday.

This post breaks down what an admin actually does at a small supply yard, module by module, and tells the truth about what AI replaces — and what it doesn't.

The reference role

Before the breakdown, the role itself. A part-time office admin at a small landscape & aggregate supply yard in Northern Kentucky / Greater Cincinnati / comparable Midwest markets, 25–30 hours/week, all-in cost:

• Wages: $18–$24/hr × ~28 hrs × 50 weeks = $25,000–$33,500/yr
• Payroll taxes (FICA, FUTA, SUTA, workers' comp): +12–15%
• Basic benefits if you offer any (no health insurance, just PTO): +$1,500/yr
• Training time, supervision, hiring/turnover overhead amortized: +$3,000–$5,000/yr
• Workspace, phone, software seats: +$1,500–$2,500/yr

All-in: $42,000–$54,000/year for a part-time role. Full-time pushes $65K–$80K. Most yards in this revenue band run part-time and the owner picks up the slack. Base wage figures derived from BLS Occupational Employment and Wage Statistics (OEWS) 43-6014 office clerks, Kentucky non-metro and Cincinnati-Northern KY metro, 2024 release; overhead figures are Rylo's small-business benchmarks.

Now: what does that person actually do all week?

The 12 jobs an admin actually does

From shadowing yard offices and from running operations myself, here's the honest list. I've grouped each into one of three buckets: AI replaces, AI assists, or still a human job.

1. Answering the phone — AI replaces

The single biggest time sink. 80+ calls/week, most of them asking variations of "do you have X in stock and what's a load delivered to Y." Voice AI configured against your real catalog handles 85–95% of these end-to-end: greets the caller, quotes the price, books the delivery, sends a confirmation SMS. Edge cases (angry customers, unfamiliar requests, anything off-script) escalate to the owner.

Time the human spent on this: ~10 hrs/week. What AI does better: never goes to lunch, never misses a Saturday, never fumbles a price.

2. Taking orders & writing tickets — AI replaces

Logging the order into whatever system you use (QuickBooks, a spreadsheet, a paper ticket book). AI does this automatically as part of the call flow — by the time the customer hangs up, the order is already in the system with timestamp, SKU, quantity, delivery address, and quoted price.

Time the human spent on this: ~4 hrs/week. What AI does better: zero data entry errors, zero illegible handwriting, zero "we lost the ticket."

3. Quoting custom or unusual jobs — AI assists

Standard catalog items, AI nails. But "I need 14 yards of a custom mulch blend with 30% red dyed and 70% natural double-shred, can you do that?" — that's a judgment call. AI drafts a quote based on component pricing and flags it for owner review before sending. Owner approves or adjusts in 30 seconds.

Time the human spent on this: ~2 hrs/week. What AI does: drafts the first pass instead of the owner starting from scratch.

4. Scheduling deliveries & dispatching drivers — AI replaces

Booking loads into the calendar with conflict detection, sequencing the route, sending the 6 AM driver SMS with the day's addresses, loads, and notes. AI does all of this automatically off the order data.

Time the human spent on this: ~4 hrs/week. What AI does better: conflict detection is mathematical, not "I think we already have a load going out then." Driver briefing goes out at 6 AM whether or not anyone got to the office.

5. Invoicing & A/R follow-up — AI replaces

Triggering the invoice in QuickBooks when delivery completes, sending the pay link, running Day 7 / 14 / 21 follow-up on overdue invoices. AI does all of this without human intervention. Owner only gets pinged when an invoice hits 30+ days and the friendly automated reminders haven't worked — at which point a personal phone call is the right move anyway.

Time the human spent on this: ~3 hrs/week. What AI does better: follow-up actually happens, on schedule, without anyone having to remember.

6. Handling walk-in customers at the counter — Still a human job

Walk-ins are a yard worker's job, not an office admin's, in most operations — but worth flagging. AI doesn't handle this. Doesn't try to. The yard worker who's already loading trucks is the one who points at the bag of rye seed, runs the card on the iPad, and hands over the receipt.

What AI does: nothing. Why this is fine: walk-ins are a small fraction of revenue at most yards (the volume is in deliveries) and the yard staff already covers them as part of normal operations.

7. Sending review requests & responding to reviews — AI replaces

Texting a Google review request 24 hours after delivery completes. Drafting responses to incoming reviews — good and bad — for the owner's one-tap approval before they go live.

Time the human spent on this: ~1 hr/week (and honestly, most owners just don't do it). What AI does better: consistent cadence. Review program runs on autopilot instead of being the thing that gets dropped first when the week gets busy.

8. Inventory tracking & reorder alerts — AI assists

The yard worker still updates stock levels at end of day from a phone form. AI watches for SKUs dropping below threshold and pings the owner with a reorder draft ready to send to the supplier. Human still picks up the phone with the supplier or hits send on the PO.

Time the human spent on this: ~2 hrs/week tracking, plus reactive scrambles when something runs out unexpectedly. What AI does: eliminates the reactive scrambles. Reorder happens before you stock out.

9. Payroll & basic bookkeeping — Still a human job

Cutting paychecks, reconciling QuickBooks at month-end, dealing with the CPA at tax time. These are not Rylo's job, and they're not AI's job. Either you do them yourself, your spouse handles them (very common), you have a part-time bookkeeper, or you outsource to a fractional CFO service.

What AI does: nothing meaningful here. What AI does adjacent to this: keeps QuickBooks clean and current as the source of truth, which makes whoever is doing the bookkeeping faster.

10. Vendor relationships, bulk-buy negotiations, supplier visits — Still a human job

The annual mulch supplier negotiation. The relationship with the quarry. Trade-show conversations that turn into seasonal pricing deals. This is the owner's job, period. Even at yards that have admins, owners handle this themselves because the relationships matter to the margin.

What AI does: nothing. Why that's correct: supplier relationships compound over years and can't be delegated to anything that doesn't have skin in the game.

11. The 7 AM "what do I need to know today" briefing — AI replaces

An admin who's been around a while can tell the owner, over coffee at 7 AM: yesterday's revenue, today's deliveries, overdue invoices, what's running low, what needs the owner's attention. That's a hard skill that takes 6+ months of ramp.

AI generates this same briefing automatically every morning, delivered to the owner's phone in under 180 words. No coffee required. No ramp time.

Time the human spent on this: ~30 min/day = 2.5 hrs/week. What AI does better: consistent, fact-based, doesn't have moods.

12. Dealing with the genuinely angry, weird, or complicated customer — Still a human job

The contractor whose load got dumped in the wrong driveway. The homeowner whose check bounced and is denying it. The job-site dispute over what color mulch was actually ordered. These need a human with judgment, authority, and ideally the founder on the other end of the phone.

AI's correct response to these is: flag, escalate, get out of the way.

What AI does: recognizes the situation, pulls the call out of the automated flow, and pings the owner immediately. What it doesn't do: try to resolve it.

Adding it up

Of the 12 jobs:

• 6 are fully replaced by Rylo OS - Yard — the high-volume, low-judgment work. Phone, ticket entry, scheduling, invoicing, reviews, briefing.
• 2 are assisted — AI does the first pass, human approves. Custom quoting, inventory reorders.
• 4 are still a human job — walk-ins, bookkeeping, vendor relationships, hard customer situations.

Reality check on the time math: a part-time admin spends roughly 25 of their 28 weekly hours on items 1–8 and 11. Rylo OS - Yard replaces or assists on all of those. The remaining items (walk-ins, bookkeeping, vendor relationships, hard customers) consume maybe 3 hrs of an admin's week — and at most yards, the owner ends up doing them anyway.

What this means for the hiring decision

If you're a yard owner currently weighing "hire an admin OR install Rylo OS - Yard," the honest framing is:

• An admin gives you a person. Someone to talk to, train, schedule, manage, and worry about. Covers maybe 50% of inbound call volume (no nights, weekends, lunches). All-in cost: $42K–$54K/yr.
• Rylo OS - Yard gives you a system. Covers 85–95% of inbound call volume, runs the back-office workflows on autopilot, and never quits. Cost: $3,500 install + $1,750/mo = $24,500/yr first year, $21K/yr after.

The math favors Rylo OS - Yard by a wide margin on dollars. But the more honest answer is: most yards in this band don't actually need a full admin role at all anymore. They need 6 of the 12 jobs done reliably (Rylo OS - Yard), plus 2 assisted (Rylo OS - Yard + owner), plus 4 hours of owner time on the things that legitimately need a human. Total weekly owner time on office work after install: ~4 hrs, down from the 25+ that's currently eating their week.

That's the actual deliverable. Not "AI replaces your admin." More like "AI replaces the parts of the admin role that should never have been a human's job in the first place, and frees the owner up to do the parts that should."

What we're not claiming

• That AI is always cheaper than a person. If you have a great admin you trust, who covers walk-ins and answers the phone and handles the supplier calls and is paid market rate, keep them. The math against Rylo OS - Yard isn't about firing competent people — it's about not having to hire one in the first place, or about freeing the great one you have to do higher-leverage work.
• That AI handles every situation in those 6 "replaced" categories. Edge cases escalate by design. The AI handles volume; you keep judgment.
• That this works for every business size. If your yard is over $3M revenue, you probably need a full-time office staff regardless — Rylo OS - Yard makes that staff faster, not redundant. The clean replacement math is for the $500K–$2M independent yard band.

How to think about your specific yard

Two ways to size this:

1. Run the missed-call calculator to see what just job #1 (answering the phone) is currently costing you. That's the floor on what Rylo OS - Yard recovers, before the other 5 fully-replaced modules add anything.
2. If the number stings, book a 15-minute intro call. We'll walk your actual operation against this 12-job list and tell you honestly which jobs Rylo OS - Yard replaces for you, which it assists on, and which still need a human — including whether that human should be you, your spouse, an admin, or a fractional bookkeeper.

The frame that works isn't "AI replaces your admin." It's "AI takes the volume work off your plate so the human work — yours, or someone you actually want to hire — has room to happen."