Operations
Headcount grows where packets stall, not where the org chart says back office.
Shared services, AP, onboarding, and operations support still move through email, shared drives, and the person who knows which system is lying. A back-office bot that summarizes the file does not change that cost.
21%
scaled agent use in service operations inside the technology sector. Elsewhere, scaled use is still in the single digits.
Stanford HAI, AI Index 20261
How I see it
The back office is a queue of incomplete packets
Back-office work looks administrative. The expensive part is reconstruction: finding the invoice, the approval, the missing ID, the exception note, then updating the system of record and chasing the next person. That is where volume turns into headcount.
Miami distribution, specialty finance, property operations, and professional services all run this pattern. The official process is clean. The unofficial path is a side spreadsheet and a queue of cases that are 80 percent done.
AI is useful when it completes the packet and writes back. It is theater when it drafts a summary a person still has to rebuild. Rules still belong where the logic is stable. Interpretation belongs where the document, the exception, or the next action is not a lookup.
Start with one queue that already has volume and an owner: invoice exceptions, employee or vendor onboarding, order cleanup, or a shared-services request type. Prove a conservative improvement in cycle time or cost per case. Do not launch a shared-services transformation.
Where the cost sits
Back-office surfacepeople on the queue × loaded cost × time on reconstruction and chase × conservative improvement
Use one request type. Do not average the entire shared-services function into one number.
Is this your problem?
Five signals the cost is already real.
Which back-office queue takes the most calendar time: invoices, onboarding, orders, or exceptions?
If you cannot name it, you are not ready to fund a back-office program.How many people touch a typical case before it is complete?
Handoffs are usually the cost, not the keystroke.Where does the unofficial path live: inbox, spreadsheet, or shared drive?
If the truth is not in the system of record, software cannot operate the queue.Which fields can be assembled without review, and which cannot?
If every field needs a person, you will not get leverage.Would a 25 percent reduction in cycle time or cost per case be material?
If not, pick a more expensive queue.
What done looks like
The result, not the category.
A solution is finished when a named result no longer waits on reconstruction.
Complete the packet
Assemble the missing invoice, approval, ID, or exception note so the case is ready to act.
A person no longer rebuilds the file from email and shared drives.
The system produces a summary someone still has to reconstruct.
Write it back to the system of record
The result lands in the ERP, HRIS, or operations system. A side spreadsheet is not done.
The official system is current without a paste step.
Someone still re-keys the answer.
Retire the unofficial queue
Inbox, chase, and the person who knows which system is lying stop being the path.
Exceptions have an owner and the old side path is closed.
The bot is live and the spreadsheet is still the truth.
What usually fails
How solution shopping wastes a quarter.
Automating the official path only
The exceptions are the work. If the new path cannot handle them, people will keep the old path.
Leaving the system of record to a human paste step
If someone still re-keys the result into the ERP or HRIS, you automated a side task.
Buying a back-office suite before naming the case
A platform is still a tool. Name the completed packet, the owner, and the baseline first.
This is not the problem if
Do not start with a category purchase.
- The request is a shared-services vision, not one queue with volume.
- Nobody will share exception rates or the unofficial path.
- A cleaner form or a rule would remove most of the cost.
- Every case is treated as too risky for software to assemble.
A useful next step
Bring one expensive problem. Get guided into production.
We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.
Discuss an AI opportunity