Business Automation Consultant, Jordan

Automation is not the same as integration

Integration moves data between systems: the ERP and the e-invoicing system, the ERP and a bank, the ERP and a warehouse. That is described on my system integration page for Jordan. Automation deals with the human steps that remain once data is in place: the reminders, approvals, checks, lists and reports people still produce by hand.

Typical candidates in Jordanian companies include:

• A register of post-dated checks kept in a spreadsheet and scanned each morning for anything due.
• Staff logging into the e-invoicing portal to see which invoices were accepted, then updating a list.
• Sales teams learning about short-dated stock only when a pharmacy refuses delivery.
• Project managers sending individual emails to collect timesheets before billing.
• Accounts assistants keying supplier invoices, some in Arabic, some in English.
• Monthly sell-out and stock reports for foreign principals built by copying exports into the principal's template.

Before building anything, I check that the rule behind each task is agreed. If nobody has decided who handles a rejected invoice or who may release quarantined stock, automation cannot route the work. In that case the process consulting page for Jordan comes first. Where the rule is clear, the task goes on a ranked list that management approves.

Post-dated checks and collections on autopilot, with people in charge

Where post-dated checks are a normal way to pay, the work around them is steady and rule-based, which makes it a good automation candidate. The decisions stay with credit control; the system takes over the listing, reminding and matching.

• Maturity lists: each week, a list of checks received and issued that mature in the coming days, grouped by bank account, so deposits and balances are planned.
• Deposit tracking: checks marked as deposited are compared with the bank statement once it is imported, and anything not cleared after the expected interval is flagged.
• Returned checks: recording a return triggers a task for the credit controller, a hold on the customer's further credit if your policy says so, and a note for the sales rep's next visit.
• Statements and reminders: customer statements generated on a fixed cycle, with reminders before and after due dates, written in wording your team approves in Arabic and English.

Messages to customers should respect Jordan's personal data protection law; consent, channel and content are for your legal advisor to confirm before automated sending begins. The automation keeps a record of every message sent, so a dispute can be answered with facts. Credit holds and legal follow-up remain human decisions, prepared by the system and approved by a named person.

Rejected e-invoices as a queue, not a daily hunt

Once the connection to the national e-invoicing system works, a different kind of manual work often appears: checking statuses, spotting rejections, finding out why and getting the right person to fix them. Without structure, this falls to whoever happens to notice, and a rejected invoice can sit until the customer complains or the sales tax return exposes the gap.

I design the exception handling as a workflow:

1. Submissions that are rejected or remain pending past an agreed point are collected into a single queue, using the status the connector writes back to the ERP.
2. Each item is assigned by reason: customer data to sales administration, item or tax data to the person who maintains master data, technical errors to whoever supports the connector.
3. Reminders continue until the item is corrected and resubmitted, or cancelled with an approved credit note.
4. A daily summary shows the finance manager the open items and their age.

The rules for what an invoice must contain, and when a credit note is required, come from your tax advisor; the workflow applies them and does not interpret them. The connector itself, and its retries and logs, are integration work owned by your implementer or developers. Automation makes sure the human side of the process keeps pace. For the wider finance picture, see finance automation.

Expiry warnings, timesheet reminders and reports for principals

Different parts of Jordan's economy repeat different manual work, and the automation backlog should reflect your business rather than a generic list.

Distributors of medicines, medical supplies and food benefit from alerts tied to batch data: stock nearing expiry flagged to sales while there is still time to sell or return it, quarantined batches waiting too long for release flagged to quality, and reorder warnings for lines that must never run out. A weekly report to each foreign principal, in the layout it expects, can be generated from the ERP and sent after a person reviews it.

IT, outsourcing and consulting firms gain most from reminders that keep billing on time: consultants reminded to complete timesheets, managers reminded to approve them, and finance alerted when a milestone is marked accepted so the dollar invoice goes out promptly. A weekly utilization and unbilled work report saves an evening of spreadsheet work.

Regional offices following parent or donor procurement rules can automate the file checklist: a purchase cannot move to payment until the required quotations, approvals and declarations are attached.

Each of these uses data already in your systems. Where that data is unreliable, such as batch numbers not recorded at receipt, I flag the gap first, because an alert based on wrong data is worse than no alert. Sector context is on pharmaceutical ERP and IT services ERP.

AI-assisted intake, who builds it, and what stays manual

Supplier invoices and receipts are where AI helps most in a Jordanian back office. Documents arrive in Arabic, English or both, in many layouts. An AI step can read supplier, dates, lines, tax and totals, then validation checks dinar amounts at three decimals, matches the order and receipt, and confirms the tax fields. Anything uncertain or failing a check waits for a person. Your bilingual staff confirm Arabic content; the AI output is always a draft.

Tool choice follows where the work lives. A rule inside Zoho can run as a workflow or a short Deluge function; inside Odoo, as an automation rule or scheduled action; inside ERPNext, as a workflow with notifications. Work that hops between a mailbox, a spreadsheet and the ERP may suit n8n, Make or Zapier. Within that range I set the automations up myself. When a task needs custom code, touches the e-invoicing connector or falls under your implementer's support contract, their developers build it from a written specification, and I run acceptance tests alongside your staff.

Some steps should stay manual: approving a credit hold, deciding the tax treatment of an unusual sale, releasing a payment, releasing quarantined medicine and sending a disputed statement. Automation prepares and records them.

Success is judged by what staff notice: fewer forgotten checks, rejections fixed the day they appear, timesheets complete before billing and a review queue that stays short. Each automation has an owner and a runbook. The wider method sits on my automation service page, while the Jordan hub gathers the remaining services.