A manual, error-prone ADP-to-Business Central data entry process โ done by hand every two weeks for years โ got automated in one afternoon. First with Claude. Then, once proven, with a zero-token script that runs on its own.
Don pulls departmental hours and payroll data out of ADP as a PDF, then keys every figure into an Excel template by hand before uploading the totals into Business Central, our ERP. Nothing about this process was automated โ and nothing about it needed to be this manual.
Rather than jumping straight to custom software, we used Claude to prove the concept was even possible โ with almost no setup. Once it clearly worked, we converted that proven workflow into a simple automated script, so the day-to-day version doesn't use AI or burn tokens at all.
Don uploads the biweekly hours & payroll PDF straight into Claude.
The same Business Central upload template he already used.
Every line item from the PDF is matched and entered into the correct cells.
Same final step as before โ just no manual typing in the middle.
The downloaded ADP PDF and his Excel template โ that's the only manual step left.
One click. No AI interface, no prompts, nothing new to learn.
A rules-based program reads the PDF and fills the Excel template โ deterministic, same result every time.
The proven logic runs on its own โ no ongoing AI cost for a task this repetitive.
This wasn't rolled out to a power user. It was rolled out to someone with very limited AI experience โ which is exactly why it's the right story to share.
"I used to dread this every other Friday. Now I put two files in a folder and I'm done."
The value here isn't just the hours Don got back. It's the pattern โ and how easily it repeats across the business.
An early, personal win built genuine enthusiasm for AI โ from someone who wasn't looking for it.
Claude validated the workflow in a single session before we invested in a custom script.
Once proven, deterministic tasks moved to a lightweight script โ full speed, zero ongoing token cost.
Any recurring "PDF or report โ manual entry โ spreadsheet โ upload" task across QHF is a candidate for the same two-phase approach.
If a task looks like "download a file, type numbers into another file, upload it somewhere" โ it's very likely worth five minutes of Claude's time to test.
As QHF moves from FDA-only to also being USDA-regulated, the team needed to understand what changes. Claude built the training content, Claude wrote the presenter script, and 11Labs turned it into a finished video โ with Ashley's own cloned voice narrating it.
We've operated under FDA rules for years. As we take on USDA-regulated production, the team needs a working understanding of how USDA requirements differ from what we already know. That's a training need, and it needed to exist quickly โ not in a few weeks once a video vendor could fit us in.
Each stage fed directly into the next โ Scott's existing USDA Project supplied the raw material, Claude did the research and writing, and 11Labs handled voice and video.
Scott's ongoing Claude Project for the USDA initiative already held internal reference material for the transition.
Claude pulled directly from those uploaded documents rather than starting from a blank page.
Where the internal material didn't cover it, Claude researched current FDA and USDA regulatory requirements directly.
A structured breakdown of FDA vs. USDA differences, ready to hand off โ no separate research phase needed.
As HACCP coordinator, Ashley owns rolling this training out to the team.
Specifically instructed to write in a natural, spoken presenting voice โ not a slide-reading tone.
Written slide by slide, in step with the PowerPoint Ashley already had.
A presentation-ready script, in a natural voice, without hiring outside writing help.
The PowerPoint and the Claude-written script went in together.
Ashley had previously cloned her voice in 11Labs โ no generic AI narrator needed.
Narration was matched to each slide automatically, producing a finished video with voiceover.
No recording booth, no editor, no reshoots โ and the voice sounds like Ashley because it is her voice.
This wasn't a video project handed to a production team โ it was one person, working across two AI tools, turning research into a finished training video in a single day.
"I didn't write a script from scratch or book any recording time. I gave Claude the research and asked for a natural presenter's voice, then 11Labs turned our deck into a finished video in my own voice โ the whole thing came together in a day."
The value isn't just this one video โ it's a repeatable pipeline for turning internal knowledge into finished training content.
The USDA Project folder gave Claude a running head start instead of a blank page.
Research, script, and voice/video each moved to AI โ not just one step in the middle.
Ashley's voice clone is now standing infrastructure for every future training module โ no re-recording required.
Any recurring "research โ write โ present" training need across QHF is a candidate for the same approach.
If a topic needs to go from "here's what changed" to a finished, presenter-quality training video โ it's worth testing this same pipeline before booking outside production help.
The QHF Pepperoni Pizza Bites USDA Grant of Inspection project touches 15 people across 9 workstreams, with a hard end-of-January-2027 target. Instead of one person manually rebuilding the schedule, the meeting agenda, and the hours log every week by hand, Claude runs that machinery โ and Scott spends his time on the decisions, not the data entry.
Getting QHF from FDA to USDA/FSIS jurisdiction for Pepperoni Pizza Bites means coordinating Facilities, Sanitation, HACCP, Product/Label, FSIS Application, Training, Production Validation, and Finance โ all on one schedule, all reporting to one Monday meeting. Without a system, that means one person manually reconciling status updates, rebuilding a dashboard from scratch, retyping decisions after every meeting, and chasing down labor hours in a shared spreadsheet everyone is nervous to touch.
The work started with the foundational documents, then grew into a standing operating rhythm: a weekly schedule sync, a live Monday meeting workflow with a permanent audit trail, and a labor-hours pipeline that keeps 15 people out of the spreadsheet entirely.
Scope, team, RACI, and constraints โ corrected in place as facts changed (facility location, newly surfaced contributors).
109 actions across 9 workstreams, predecessors verified, subtasks auto-normalized on every edit.
A single HTML view of active work, risks, and milestones โ rebuilt fresh every sync, not hand-maintained.
Maps which existing FDA documentation reuses, modifies, or needs to be built new under USDA/FSIS rules.
His manual edits are the one input Claude never overwrites โ conflicts get flagged, not silently resolved.
Cell-for-cell sync with formatting preserved, then normalized: subtasks detected and relabeled automatically.
Active tasks, overdue items, completions, milestones, and risks all recomputed from the actual data, plus that week's hours worked.
Typos, blank owners, orphaned tasks, and unexplained status regressions get surfaced โ never auto-fixed silently.
Dashboard's Agenda tab shows active tasks, this week's completions, hours worked, and top risks โ computed live, not typed up.
Status changes, comments, and reassignments captured right on the dashboard during the meeting.
A script applies status changes to WORKING and archives every comment into a permanent Meeting Notes Log โ never overwritten.
A dated, read-only copy of that week's dashboard is archived to SharePoint โ a permanent record no live edit can ever change.
15 team members log hours via dropdown โ name, pay type, hours, date, workstream, description.
Every submission writes straight into the LaborLog Excel table โ no one opens the spreadsheet directly.
The Monday agenda's "Hours Worked Last Week" section reads the log live and groups it by person automatically.
The table structure and validation lists stay intact because no one is hand-editing the underlying file.
The recurring machinery is only half of it. Day to day, Claude is also the first stop for the tactical work that used to eat hours between meetings โ the kind of work that never justifies its own custom system, but still has to get done.
Drafted full CARs for equipment purchases (e.g., the Oven 1 infeed covers, the EnSURE Touch ATP monitor) โ ROI, capital requirements, and success criteria ready for sign-off.
FSIS requirements, lighting and facility standards, chemical approvals โ pulled together and cited on demand instead of hours of independent digging.
Vendor outreach, team status updates, and follow-ups drafted in Scott's voice, ready for a quick review pass before sending.
Legacy Word/PDF sanitation SOPs and FDA-era documentation reformatted and restructured to match what USDA/FSIS actually requires.
This isn't a one-time build. It's an ongoing operating system for a regulatory project with a hard deadline, run by one project manager who still owns every decision โ just none of the retyping.
"There's no way I could run a 109-action regulatory project like this on top of everything else on my desk as CEO without Claude. I don't have a PM analyst to hand this to โ Claude is the reason I can still be the one running it myself."
The value isn't any single document Claude built. It's that the system now runs the recurring, error-prone parts of project management on its own โ while the PM keeps full control over every decision.
Charter, schedule, and dashboard came before any automation โ the system only works because the underlying data structure is right.
Claude flags conflicts and data-quality issues rather than silently resolving them โ the PM still decides.
A Form instead of a shared spreadsheet means 15 people can contribute without any risk of breaking the file.
Every meeting gets a frozen, dated snapshot and a permanent comment history โ nothing is ever silently lost.
Any initiative with multiple workstreams, a recurring status meeting, and a shared tracking file is a candidate for this same layered approach โ foundation documents, then a weekly sync, then a source-of-truth data pipeline.
| Operational DT | 4.1% | PD waste (yest) | 5.0% |
| Mechanical DT | 3.1% | WTD waste | 5.8% |
| Unplanned Maintenance | 28 |
| Drive fault | 18 |
| Sensor fault | 10 |
| Unplanned Operations | 15 |
| Changeover overrun | 15 |
| Planned Operations | 9 |
| Sanitation break | 9 |
| Operational DT | 5.2% | PD waste (yest) | 6.5% |
| Mechanical DT | 4.3% | WTD waste | 7.0% |
| Unplanned Maintenance | 40 |
| Belt tracking | 25 |
| Oven trip | 15 |
| Unplanned Operations | 18 |
| Product jam | 18 |
| Planned Operations | 10 |
| Shift changeover | 10 |
| Tier | Item | Description | Customer | Cs/hr | CM% | CM/hr | Illustrative YTD Rev |
|---|---|---|---|---|---|---|---|
| Tier 1 โ โฅ$6,000/hr | |||||||
| T1 | SKU-1000 | Filled Pizza Crust Bites โ Jalapeรฑo (New Product) | Customer A | 352 | 74.0% | $8,080 | $8,452 |
| T1 | SKU-1001 | Mozzarella Stuffed Bites โ Garlic | Customer D | 450 | 52.0% | $8,640 | $63,762 |
| T1 | SKU-1002 | Filled Pizza Crust Bites โ Garlic (New Product) | Customer A | 352 | 63.3% | $7,870 | $27,841 |
| T1 | SKU-1003 | Mozzarella Pull-Apart โ Bulk | Customer D | 350 | 41.5% | $5,970 | $245,601 |
| T1 | SKU-1004 | Mozzarella Breadstick โ Bulk | Customer E | 343 | 44.6% | $6,330 | $341,698 |
| T1 | SKU-1005 | Cheesy Pull-Apart โ Sweet | Customer A | 420 | 35.2% | $6,460 | $171,216 |
| T1 | SKU-1006 | Mozzarella Bites โ Jalapeรฑo | Customer B | 338 | 36.5% | $5,690 | $16,016 |
| T1 | SKU-1007 | Mozzarella Bites โ Plain | Customer B | 338 | 37.6% | $6,030 | $18,591 |
| Tier 2 โ $5,000โ5,999 | |||||||
| T2 | SKU-1008 | Mozzarella Bites โ Classic | Customer D | 450 | 37.8% | $5,170 | $40,712 |
| T2 | SKU-1009 | Mozzarella Bites โ Standard | Customer F | 350 | 39.5% | $5,270 | $294,941 |
| T2 | SKU-1010 | Cinnamon Mini Roll โ Semi | Customer A | 651 | 41.9% | $5,390 | $233,380 |
| T2 | SKU-1011 | Strawberry Mini Bagel โ Semi | Customer A | 651 | 41.2% | $4,790 | $247,369 |
| T2 | SKU-1012 | Cinnamon Mini Bagel โ Retail | Customer A | 651 | 41.5% | $5,080 | $82,503 |
| T2 | SKU-1013 | Garlic Cheesy Pull-Apart | Customer A | 420 | 30.1% | $5,320 | $2,321,963 |
| T2 | SKU-1014 | Mini Cinnamon Roll โ Semi (2) | Customer A | 650 | 46.0% | $4,760 | $236,655 |
| T2 | SKU-1015 | Strawberry Mini Bagel โ Retail | Customer A | 651 | 39.6% | $4,990 | $241,778 |
| T2 | SKU-1016 | Mini Cinnamon Roll โ Foodservice | Customer A | 650 | 45.2% | $4,440 | $3,433,469 |
| T2 | SKU-1017 | Mini Cinnamon Roll โ Caramel Retail | Customer A | 650 | 42.1% | $4,750 | $8,292 |
| Tier 3 โ $4,500โ4,999 | |||||||
| T3 | SKU-1018 | Stuffed Cinnamon Churro Bite | Customer G | 240 | 64.4% | $5,030 | $36,138 |
| T3 | SKU-1019 | Grilled Cheese Pull-Apart โ Bulk | Customer B | 197 | 37.2% | $4,490 | $25,396 |
| T3 | SKU-1020 | Mini Cinnamon Roll โ Retail | Customer A | 650 | 44.9% | $4,800 | $497,963 |
| T3 | SKU-1021 | Grilled Cheese Pull-Apart โ IW | Customer B | 197 | 35.5% | $5,090 | $26,217 |
| T3 | SKU-1022 | Cinnamon Mini Bagel โ Foodservice | Customer A | 651 | 39.0% | $4,540 | $820,445 |
| T3 | SKU-1023 | Mini Cinnamon Roll โ Caramel Foodservice | Customer A | 650 | 39.5% | $4,790 | $497,525 |
| T3 | SKU-1024 | Filled Pastry Stick โ Pumpkin Bulk | Customer I | 338 | 54.7% | $5,080 | $6,600 |
| T3 | SKU-1025 | Filled Pastry Stick โ Pumpkin IW | Customer I | 338 | 52.4% | $4,560 | $13,523 |
| T3 | SKU-1026 | Stuffed Pizza Bites | Customer B | 238 | 32.1% | $4,570 | $555,564 |
| T3 | SKU-1027 | Mozzarella Pull-Apart โ Bulk (2) | Customer C | 233 | 59.9% | $4,040 | $103,378 |
| Watch โ <$4,500 | |||||||
| WX | SKU-1028 | Strawberry Mini Bagel โ Foodservice | Customer A | 651 | 35.8% | $4,290 | $2,917,147 |
| WX | SKU-1029 | Stuffed Chocolate Churro Bite | Customer G | 240 | 56.8% | $4,480 | $30,586 |
| WX | SKU-1030 | Mozzarella Pull-Apart โ IW | Customer C | 233 | 57.2% | $3,990 | $351,700 |
| WX | SKU-1031 | Filled Crescent โ Grape | Customer A | 500 | 42.5% | $4,130 | $60,818 |
| WX | SKU-1032 | Filled Crescent โ Chocolate | Customer A | 500 | 44.9% | $4,400 | $263,361 |
| WX | SKU-1033 | Sweet Cream Filled Roll โ IW | Customer B | 262 | 39.3% | $3,740 | $100,922 |
| WX | SKU-1034 | Filled Pastry Stick โ Strawberry & Cream | Customer I | 338 | 41.9% | $3,830 | $30,683 |
| WX | SKU-1035 | Filled Pastry โ Apple | Customer A | 651 | 37.0% | $3,410 | $1,234,857 |
| WX | SKU-1036 | Filled Pastry โ Cherry | Customer A | 651 | 30.5% | $2,970 | $686,087 |
| WX | SKU-1037 | Cereal-Coated Filled Pastry | Customer A | 458 | 35.2% | $3,640 | $1,246,334 |
| WX | SKU-1038 | Stuffed Pizza Roll โ Bulk | Customer B | 250 | 23.9% | $3,190 | $238,697 |
| WX | SKU-1039 | Stuffed Pizza Roll โ IW | Customer B | 250 | 21.8% | $3,240 | $404,606 |
| WX | SKU-1040 | Garlic Breadstick โ 7 inch | Customer H | 162 | 53.7% | $3,370 | $37,665 |
Fill out the form below, and one of our team members will reach out to discuss your companies needs.