Zero-Rekey Intake: email and PDFs become a job record.
Zero-Rekey Intake reads incoming email and PDFs. Confirmed fields write into the job system. Anything uncertain stops for a person. The job system stays the system of record.
Confirmed fields pack into a job. Uncertain fields peel off.
A customer email, a PDF, a photo of a docket. Data entry automation reads it, fills the fields the job system needs, and writes the record. Missing site, an unmatched customer, or a clash with a job that already exists stops for a person, with the source attached.
The order is retyped from the inbox. The draft has to live in the book.
Most order desks still retype the inbox into the book. Mail arrives, someone opens the PDF, then they type the same customer, site, product, and date into the system they already pay for.
What actually arrives is mail. Sometimes a PDF or a photo. Sometimes a website form. The form is treated as the customer, not the mailbox that forwarded it.
The first pass only asks whether this looks like an order, a collection, or something else. The real decision waits until the whole thread is in front of it. A first message becomes a new order. A later message is a new order, an update to an open draft, or noise. A thank-you is left alone. The same mail twice is not written twice.
Then it looks things up. The rest of the conversation. The attachment. The customer in the book. Opening hours and whether a truck can get in. What that customer has actually taken before, but only from confirmed work.
The attachment wins on lines and on the delivery address. Product has to be a live catalogue item. It will not invent one, and it will not copy last time's PO onto this one.
What gets written is a draft in the book. Not a live job. Each field keeps a note of where it came from. A reply is left sitting as a draft. Nothing goes out.
It stops for a person when the street is missing, when the address is a PO box, when the attachment and the book disagree, when the customer is ambiguous, or when someone replies on an order that is already live. Confirm is still a person.
The point is not a smarter inbox. The point is that the retype is gone, and the only work left is the exception.
How Zero-Rekey Intake runs data entry automation.
Zero-rekey does not mean no human ever touches the work. It means a person does not re-type fields the source already contained, on the normal path, into the system of record.
Email in
The mailbox, a PDF attachment, a photo of a docket, or a body that is the order. Keep the file. The source is evidence, not a prompt to discard.
Read the fields
Read the document and fill the fields the job system needs: customer, site, requested date, lines, notes. Match them to records you already hold. Do not invent a customer.
Stop if uncertain
Missing required fields, an unmatched customer or site, a clash with an existing job, or a write you cannot cite. Those cases go to a person with the source attached.
Write the job
Confirmed fields create or update the record in the job system you already run. Your schema, your status codes, your book. A second list of jobs to one side is still re-keying.
The coordinator is not employed to type. They are employed to get jobs into the book correctly so the field team can move. The tax is the seam: unstructured inbound, structured job record.
Cases that fail a stop condition go to a person with that evidence assembled. That is the design, not a failure of intake. The human's job becomes judgment and missing information, not typing the obvious lines again.
Identify the object. Then fill the schema.
This page is the use-case. The method is two agents, thread tools, and the schema that starts the workflow. One client-reported desk: over 1,000 emails a day across inboxes, and 2–3 hours a day back across departments. Not one FTE deleted. Not a guarantee.
When Zero-Rekey Intake cannot confirm, it stops.
Missing required fields, an unmatched customer or site, a clash with an existing job, or a write you cannot cite. Those cases go to the desk with the email or PDF still attached.
Done means the job exists in the book.
Confirmed fields create or update the record in the job system you already run. Your schema, your status codes, your book. A second list of jobs to one side is still re-keying.
Intake software that keeps its own list of jobs is a second book. Dispatch, invoicing, the field app, and month-end will not use it. The workflow is finished when the job system has the record, in the shape the rest of the operation already trusts.
A chatbot that summarises the email has not closed the seam. PDF intake that does not create the job has not closed the seam. The write into the job system is the work.
This is not invoice capture.
Keep the seam boring so the owner, the system of record, and the exception path stay honest.
In
Channels that create or change a job: a sales email, a customer PDF, a supplier pack that is actually a work instruction. The finish line is a job, work order, or job-management record.
Out
Supplier invoices for payment, credit notes, remittance, three-way match, posting to the ledger. Different owner. Different controls. Do not buy one product and expect it to be both.
Same seam. Different floor.
The details change. The method does not: read the source, fill what you can confirm, stop when you cannot, write the job in the system of record.
The order is already in the inbox.
A customer or sales email with a PDF, or a body that is the order. The coordinator opens the mailbox, opens the job screen, and types. Missing site, an unknown item, or "as per last time" stays with a person. The job still has to land in the book before anyone can start.
Instructions that should already be a job.
Booking instructions, packing lists, and customer PDFs that should create a job in the book. The work on this page is getting the record in. It is not chasing ETAs, and it is not matching carrier invoices for payment.
The callout has to become a job first.
A request email, a photo of a docket, a customer note that should open a job. The person on the desk is there to get the job in correctly so the field team can move. Typing the lines the source already contained is the tax, not the job.
Measure the seam. Then write jobs, not summaries.
Workflow Launch is the commercial home for this intake workflow. Workflow Scoping is the measurement step if you do not yet have a signed baseline.
Workflow Launch
One named intake workflow, connected to the mailbox and the job system you already run, with a bounded normal path, an exception queue, quality checks, source records, a capacity dashboard, training, and 30-day stabilisation. From $14,995 AUD ex GST per workflow, including all scoping expenses. Bigger builds cost more.
Workflow Scoping
Map the mailbox with the people who currently sit it. Set the KPI and evidence method. Freeze the stop conditions. From $2,995 AUD ex GST per workflow. If the same workflow proceeds to Launch, scoping is not charged on top.
After go-live, AI Accelerator keeps the accepted system supported from $995/m AUD ex GST per workflow. New features are quoted separately. This page is one of the patterns on common bottlenecks. Once the job exists, job dispatch and scheduling is the next seam. Turning a request into a reviewed draft is quote preparation. How we evaluate a production build partner lives on AI automation agency. The measurement step is an AI readiness assessment.
Bring the mailbox, not a technology brief.
Workflow Launch is the commercial home for this intake workflow. If the seam is not yet measured, start with Workflow Scoping.