ENID
Digital Development

Mari Rekap AI Document Automation

An AI web app that turns receipts, invoices, PDFs and chat messages into clean, reviewable tables for Indonesian businesses, with every figure linked to its source document and ready to export to Excel.

Year : 2026
Industry : Software as a Service (SaaS)
Mari Rekap homepage showing a receipt being turned into a data table

Mari Rekap is the first product of MKL (Mari Kita Lembur). It was built for small and mid-sized businesses in Indonesia that still retype receipts, invoices, purchase orders and payment confirmations from WhatsApp into spreadsheets by hand. That work is slow, a single typo means rechecking every row, and when someone asks where a number came from, the proof is buried in a phone gallery or a chat thread.

Custompedia designed and built Mari Rekap end to end: the product, the public website at rekap.marikitalembur.com, the web app and the document-processing pipeline behind it.

Objectives

• Remove manual data entry from the recap work of small business teams.

• Let users send photos, PDFs and pasted chat text in bulk, from a phone or a laptop.

• Keep people in control: every extracted figure is reviewed next to its source before it enters the table.

• Keep the proof: each row stays linked to the document it came from.

• Launch as a self-serve product with a free start and paid monthly plans.

Approach

The insight was that the hard part is not reading a document but trusting the number. So Mari Rekap is built around review, not blind automation.

Users upload up to 100 files at once, take a photo of a receipt or paste text from a chat. AI reads each document into the user's own table format, or suggests a table when they do not have one yet, and the system checks that the figures agree: subtotal, tax and total. Results appear side by side with the original document, with a confidence level, so the team only checks what is flagged before approving it.

Approved rows land in the Transactions table with their proof attached, and the whole table exports to Excel at any time. Owners, admins and editors can work in the same workspace. The app runs on Cloudflare Workers with D1, R2 storage and queues, so large batches keep processing in the background without making the user wait.

The marketing site explains the flow in plain Indonesian, with an animated teaser that follows a receipt from upload to Excel.

Result

100 files Uploaded and processed in a single batch
3 inputs File upload, phone camera and pasted chat text
100% Of approved rows stay linked to their source document

Explore More Projects

LET’S CREATE SOMETHING AMAZING

GET IN TOUCH