How MooBerry Farms cut
weekly inventory from
2.5 hours to 31 minutes
with an AI Voice Agent.
Ryan replaced a clipboard, his mother-in-law's keyboard, and four steps of double data entry with Betty, a custom voice agent who takes inventory by phone, confirms every number out loud, and writes the whole thing to Google Sheets and Square automatically.
The receipts, before the story.
2 hrs/week × 52 weeks. Ryan gets two and a half work weeks back.
// Per session: 2.5h to 31 min. From freezer clipboard + retyping to one phone call.
90 min in the freezer + 60 min retyping, down to one phone call. First run, no practice.
Ryan and Cory walk through the first call, on the floor of MooBerry Farms.
// Ryan, owner of MooBerry, on inventory day. The clip cuts between his old paper workflow and his first call with Betty.
She sounds like my banker, which I can't really get over.
Who, what, and how long.
- Client
- MooBerry Farms
- Industry
- Direct-to-consumer cattle farm
- The Ask
- Eliminate the 2.5-hour weekly inventory bottleneck.
- What We Built
- A custom voice agent ("Betty") that takes inventory by phone and syncs to Google Sheets + Square POS.
- Stack
- ElevenLabsn8nGoogle SheetsSquare POSEmail
- Timeline
- A few weekends
A clipboard, a mother-in-law,
and 2.5 hours every week.
Ryan runs a family cattle farm in Tennessee. He breaks down whole cows into thirteen different cuts and sells direct through a farm store and an online Square shop. The product is excellent. The back office, until recently, was a brick wall.
Every week, inventory went like this. Ryan stood in the freezer and weighed every cut, one by one, reading each weight out loud. A second person wrote every number down on whatever paper was nearby: sometimes a clipboard, sometimes the back of a napkin, sometimes the back of an envelope.
Then that paper went to his mother-in-law. She typed every weight into an Excel spreadsheet. That spreadsheet fed Square POS, which generated the purchase order. Three people. Four steps. The same numbers, touched twice.
“We always gravitate back towards paper and pencil. That's how we rock. We count, we write down, we go to a computer, we type it again into a spreadsheet.”
- Same numbers entered twice: once on paper, once in Excel.
- Three people locked into the workflow every single week.
- Yield % unknown for a day or more after inventory.
- A single typo could silently corrupt Square inventory.
- Workflow broke entirely whenever someone got sick.
2.5 hrs every week. Three people. Half a workday gone, every single week, before anything ever made it onto the storefront.
Meet Betty, the voice agent who answers MooBerry's phone.
Betty lives behind a single phone number. She answers in a warm, slightly Southern voice. She is trained specifically for the way Ryan does inventory: meat types, box numbers, weights. She keeps up with him as fast as he can talk.
When Ryan finishes a box, Betty repeats the count, the total weight, and asks if it is correct. When he finishes the session, she gives him a grand total and tells him she's pushing everything to Google Sheets. There is no app. There is no spreadsheet to open. There is no second person typing.
Hear Betty answer the phone.
A short clip from Ryan's first inventory call. Listen to how Betty handles the dictation, confirms the box, and catches the 0.80-vs-80 mistake in real time.
// Audio uploads soon. Player is wired and ready.
Four steps Ryan sees.
Everything else, in the background.
Ryan calls Betty.
One phone number. Speaker on, freezer door open. Betty picks up: "MooBerry Farms, this is Betty. Are you ready to record some inventory?"
Ryan dictates a box.
Meat type, box number, then every weight off the scale. As fast as he can talk. Betty keeps up. No app, no form, no fields.
Betty confirms.
After each box: "That's 6 ribeyes, 4.81 pounds total. Sound right?" She catches mishears in real time. Ryan corrects on the call.
Betty closes out.
When inventory is done: "Total weight recorded today is 183.58 pounds. I'll send this to the Google Sheet now." That is the whole workflow.
Four things Betty is doing while Ryan talks.
- Every weight written to Ryan's existing Google Sheet, same shape, same columns, untouched.
- Summary email to Ryan with totals by box, by cut, and a grand total for the session.
- New inventory pushed to Square POS so the online storefront matches the freezer.
- Full transcript saved for every call. Error-traceable, audit-friendly.
Ryan said “0.80 pounds.”
Betty heard “80 pounds.”
Halfway through the first call, Ryan asked Betty to add one more chuck eye steak to the previous box. He said the weight was “0.80 pounds.” She heard “80 pounds.”
On the old paper system, nobody would have caught it. The error would have rolled through the spreadsheet, into Square, and onto the storefront. Customers would have seen impossible inventory levels for days.
Instead, Ryan looked at the running total Betty was building, saw the number was wrong, and corrected her in real time. She updated the entry on the spot. The rest of the inventory kept moving forward, untouched.
We corrected it in real time, and it didn't throw off the rest of the spreadsheet. That was incredible.
The full flow, start to finish.
One phone call in. Three systems updated. The dashboard MSB operates on Ryan's behalf: every call's status, impact, and history.
Betty · Inventory Voice Agent
Every metric, side by side.
| // Metric | Before | After | Delta |
|---|---|---|---|
| Time per inventory session | ~2.5 hours | ~31 minutes | first try |
| People required | 3 | 1 | −2 people |
| Steps in workflow | 4 | 1 | one call |
| Time until yield % is known | A day or more | Within 15 minutes | same evening |
| Risk of typo corrupting Square | High · silent | Self-correcting | caught on the call |
| Workflow if someone is sick | Stops entirely | Unaffected | 24/7 |
Inexpensive to run.
Built to last.
No enterprise software. No SaaS sprawl. Just one phone number that does the work of several people, running on tools that stay cheap to operate at the scale of a small farm.
Custom persona (“Betty”), tuned for noisy freezer rooms and cooler-fan audio.
Turns each transcript into strict JSON. Normalizes spoken numbers and reconciles every box, cut, and weight into clean totals.
Parses every transcript. Extracts cuts, boxes, and weights. Routes structured data.
Ryan's existing sheet, same shape, same columns. Nothing on his side had to change.
Summary with totals by box, by cut, and a grand total. Delivered within minutes of call end.
Pushes new counts to the storefront. Online shop always matches the freezer.
Self-correcting during the call. Full transcript backup of every conversation.
Cory, with My Sick Builds, came to us. He's a customer of ours. He came over to the store one day and said, where are your biggest pain points? What do you do repetitively that you don't like doing?
Every week, it's the inventorying of meats. He built Betty to receive our inventory and eliminate a pain point. The worst part of our job, other than snowstorms. He fixed that problem.
She always shows up for work. She doesn't get sick. It can be 10 o'clock at night and she's going to answer that phone.
“Keep your skepticism, but don't let it stop you from buying back time. Use technology to save you time so you can do things you enjoy more.”
Frequently asked questions
How much time did the voice agent save MooBerry Farms?
Weekly inventory went from about 2.5 hours to 31 minutes on the first run, which recovers roughly 104 hours a year. The work also went from three people down to one.
How does the inventory voice agent work?
Ryan calls one phone number and reads each weight off the scale. Betty, the voice agent, confirms every box out loud, then writes the data straight to his existing Google Sheet and syncs it to Square POS. There is no app and no second person retyping numbers.
Can the voice agent catch mistakes during the call?
Yes. On the first call Betty heard a chuck eye steak as 80 pounds when Ryan said 0.80 pounds. He saw the running total was wrong and corrected it in real time, before the error could reach the spreadsheet or the storefront.
What does it cost to run a voice agent like this?
It runs on usage-based tools (ElevenLabs for voice, n8n for orchestration, Google Sheets and the Square API) that stay inexpensive at the scale of a small farm. There is no enterprise software and no monthly seat licenses.
Have something in your week
that feels like this?
We start with a 15-minute call. We look at the parts of your week you would gladly hand off, and tell you which ones are actually worth automating.
No pitch deck. No sales call. Just the diagnosis.