REDA One Help Center
EN
Language
English EN Español ES Português PT Italiano coming soon Français FR Deutsch coming soon العربية coming soon
reda.one ↗
Home / REDA Financials AI / Administration / Batch Operations / Scheduling the OCR Processing Batch

OCR Processing Batch

The admin runbook for document intake. Unlike its daily siblings, this is a queue processor: started once, cycling every 15 minutes, turning inbox documents into draft invoices.

Getting there. Open the App Launcher (the grid icon at the top-left of any page) → REDA Financials Admin → the Financial Admin Console tab → the Operations tab, then find the OCR Document Processing card.
Access you'll need. Starting and stopping this processor requires the Accounting Admin v2 permission set; the Accounting Controller v2 can reach the console too. The Accounts Payable v2 and Accounts Receivable v2 sets cannot see it — everyone else meets only the draft invoices it creates.
The card says: “Automatically processes incoming documents through AI extraction and creates purchase invoices. Runs every 15 minutes when active.”

How this card differs from the others

The other ten cardsThis card
Pick a daily run time with ScheduleStart / Stop — a switch, not a time picker
One pass per dayCycles every 15 minutes while active
Show Next Scheduled Run / Last RunShows live queue counts — pending | extracting | errored
Run Now for a one-off passRun Now forces an immediate cycle without waiting for the next quarter-hour
Consume nothing but timeEach processed document draws on the plan's monthly AI document allowance
The expanded OCR Document Processing card. Unlike the scheduled batches it carries a DOCUMENT QUEUE strip with three counters — PENDING 0 in blue, EXTRACTING 0 in amber and ERRORS 0 in red — and instead of a Run Time it offers a Start button that begins a processing loop which reschedules itself every 15 minutes, beside a Manual Run panel whose Run Now button executes a single cycle without starting the loop.
One of the two cards that breaks the pattern. There is no run time here because OCR is a loop, not an appointment: Start sets it going and it re-schedules itself every 15 minutes. The queue counters — PENDING, EXTRACTING, ERRORS — are how you read its health at a glance.

What each 15-minute cycle does

Takes pending documents from the OCR Inbox

Everything that arrived since the last cycle — emailed invoices, uploads — waits there as pending.

Runs AI extraction

Vendor, dates, amounts, line items — read off the document and matched to your master data.

Creates draft purchase invoices

Ready for a human reviewer — extraction never posts anything by itself. Errored documents stay in the inbox with their reasons, ready for reprocessing after fixes.

Scheduling it

Not a time picker — Start it and leave it running; Stop pauses new cycles (in-flight work completes). Most orgs run it permanently: 15-minute intake latency is the feature.

A worked quarter-hour — Meridian's inbox

A vendor emails a PDF invoice to Meridian's intake address at 9:03. It lands in the OCR Inbox as pending — the card's counter ticks up by one. At the next cycle the document moves to extracting; the AI reads the vendor, dates, amounts and lines, matches them against Meridian's master data, and a draft purchase invoice appears for the AP team to review — before the coffee's done, as the deep-dive puts it. One month-end, every document suddenly errors at once: the monthly AI document allowance is exhausted, which the Subscription tab's Document Consumption card confirms in one glance. The queue holds safely until the new month's allowance opens, then drains on its own.

Read the three counters as a pipeline. Pending steadily draining is healthy. Pending growing while extracting sits at zero means the processor is stopped. Errored climbing means documents need attention in the inbox — each carries its reason.

Common mistakes to avoid

MistakeWhy it hurtsDo this instead
Looking for a daily time picker on this cardThere isn't one — it's the console's one queue processorStart it once; it cycles every 15 minutes while active
Stopping the processor to "save credits"Documents pile up as pending; the credits get spent anyway when you restart, just later and all at onceManage the allowance on the Subscription tab; leave intake flowing
Treating errored documents as lostThey're waiting, visibly, in the OCR Inbox with reasonsFix the cause (master data, document quality) and reprocess from the inbox
Ignoring the monthly allowance until everything errorsIntake halts at the busiest moment — month-endGlance at Document Consumption on the Subscription tab as part of the morning check
Expecting the batch to post invoicesIt creates drafts — review is a human step by designBuild the reviewer workflow around the OCR review screen
Stop is graceful. Stopping the processor pauses new cycles — work already in flight finishes cleanly. Nothing is dropped mid-extraction, so you can stop and start without fearing half-read documents.

Monitoring it

Green means: the card shows Scheduled, the last-run timestamp is from the expected window, and the outcome icon is informational. Anything else, click View Audit Log — the run’s record lists processed / succeeded / failed with a reason per failure. The card’s queue counters are the health check: pending / extracting / errored. Pending steadily draining is healthy; errored documents wait in the OCR Inbox with reasons, ready for reprocessing. Remember each processed document draws on the plan’s AI document allowance — the Subscription tab shows the month’s meter.

Troubleshooting

SymptomLikely cause → action
Queue growing, nothing processingThe processor is stopped — Start it. If started and stuck, check the last audit entry for the error.
Everything erroring at onceUsually the document cap: the monthly AI allowance is exhausted. Confirm on the Subscription tab.

Where to go next

  1. OCR Processing Batch — what the batch does, in depth.
  2. Operations Tab — the console.
  3. Audit Logging — reading run history.

Frequently asked questions

Because intake is continuous — vendors email invoices all day, and 15-minute freshness is the point. A nightly pass would turn same-day drafts into next-day drafts. It is one of two cards on the Operations tab that run as Start/Stop loops rather than at a set time; the other is E-Invoice Status Poll, for the same reason on the outbound side.

Check the processor is started — that's the cause almost every time. Started and still stuck: read the last audit entry for the error.

Usually the document cap — the monthly AI allowance is exhausted. Confirm on the Subscription tab; the queue holds until the allowance renews or the plan grows.

Consumption is per processed document, drawing on the plan's monthly allowance — an idle cycle with nothing pending processes nothing.

No — it creates draft purchase invoices for human review. The reviewer confirms the extraction and posts; see Creating an Invoice from OCR.

Was this article helpful?