How Edrak automated 80% of repetitive procurement work while keeping buyers in control

The company

The company operates three manufacturing facilities in Saudi Arabia and purchases thousands of production components, maintenance supplies and operational materials every year.

Its procurement team manages requests from engineering, maintenance, production and corporate departments. Each request can involve multiple supplier quotations, technical requirements, commercial comparisons and several layers of approval.

The company had already invested heavily in automating its factories. Procurement, however, still depended on email, spreadsheets and manual follow-up.

Procurement had become an information problem

The team’s buyers were not spending most of their time negotiating with suppliers or improving commercial terms.

They were collecting information.

A typical purchase request required a buyer to:

• Check whether the request contained the necessary specifications
• Search previous orders for prices and supplier history
• Contact approved suppliers for quotations
• Extract commercial terms from differently formatted documents
• Build a comparison spreadsheet
• Identify missing documents or conflicting information
• Prepare a recommendation for approval
• Follow up with the next person in the process

None of these tasks was individually difficult. Together, they consumed much of the team’s working week.

Information was distributed across email threads, ERP records, supplier quotations, technical documents and shared folders. Buyers often knew that relevant information existed, but finding and assembling it took time.

Urgent requests made the problem worse. When production needed a replacement component quickly, the procurement team had to move faster without weakening the company’s approval and compliance requirements.

Starting with one complete workflow

The company deployed Atlas around one specific workflow: converting an approved purchase request into a complete supplier comparison and approval package.

Foundry connected the company’s procurement policies, supplier records, historical purchase orders and document repositories.

The team then defined how a compliant procurement process should run:

• 1What information every purchase request must contain
• 2Which suppliers could be contacted for each category
• 3Which commercial and technical fields had to be compared
• 4When exceptions required employee review
• 5Which approvals were required at different purchase values

Atlas used these rules to coordinate the work around each request.

The underlying process did not change. Responsibility for purchasing decisions remained with the company’s employees. What changed was the amount of administrative work required to reach those decisions.

From purchase request to decision-ready package

When a new request entered the workflow, Atlas checked it against the company’s procurement requirements.

If a specification, delivery date or cost centre was missing, Atlas identified the gap before the request reached a buyer. Complete requests moved forward without waiting for manual review.

Atlas then retrieved relevant information from previous purchases held in Foundry, including:

• Suppliers previously used for the category
• Historical prices and order volumes
• Typical delivery times
• Previous commercial terms
• Available supplier performance records

Once quotations were received, Atlas extracted the relevant information regardless of document format. It compared unit prices, lead times, payment terms, warranties and technical exceptions in a consistent structure.

The system then prepared a decision package containing the original request, supplier quotations, commercial comparison, relevant purchasing history and any exceptions requiring attention.

The buyer received the completed package inside the mission room.

They could inspect the supporting source for every extracted figure, ask for additional analysis and correct information when necessary. Atlas did not select the supplier or approve the purchase. It prepared the work required for an employee to make that decision.

Automation without removing accountability

The company’s leadership did not want purchasing decisions delegated to an opaque system.

Atlas was therefore designed around explicit approval points.

Employees remained responsible for:

• Selecting the supplier
• Approving commercial exceptions
• Accepting alternative technical specifications
• Authorising expenditure
• Releasing the final purchase order

Atlas handled the work surrounding those decisions: retrieving records, reading documents, checking requirements, preparing comparisons and moving approved work to the next stage.

Every action was recorded. Buyers could see which documents had been used, where each value came from and what the system had done.

When information was uncertain or a request fell outside the company’s rules, Atlas stopped the workflow and brought the relevant employee in.

The result was not a fully autonomous procurement department. It was a procurement team with significantly less administrative work and clearer control over every purchase.

Measurable improvements within eight weeks

The company began with one purchasing category and a small group of buyers. After validating extraction accuracy and approval controls, it expanded the workflow across indirect procurement.

Within eight weeks, the team reported:

• 80% of repetitive preparation work automated, including document extraction, policy checks, historical retrieval and comparison-table preparation
• 14 hours returned per buyer each week, previously spent searching for records and assembling approval materials
• 2.4× faster processing, measured from a complete purchase request to an approved purchase order
• 63% fewer requests returned for missing information, as incomplete submissions were identified earlier
• 100% human approval retained for supplier selection, exceptions and expenditure
• Full traceability between each recommendation and its supporting source documents

The gains came primarily from eliminating the waiting and preparation between decisions, not from asking employees to make decisions faster.

More time for the work that changes the outcome

With less time spent preparing files, buyers could spend more time negotiating pricing, consolidating orders and working with suppliers.

Category managers gained a more consistent view of purchasing activity across facilities. Approval teams received standardized decision packages instead of working through different spreadsheets and email threads for every request.

The company also began capturing how its strongest buyers evaluated quotations and handled exceptions. Those decisions could be translated into repeatable workflow rules without removing the judgment of the people responsible.

Expanding from one workflow

Following the initial deployment, the company began extending Atlas into adjacent procurement activities:

• Supplier onboarding and document validation
• Contract obligation tracking
• Purchase-order discrepancy resolution
• Tender and quotation analysis
• Recurring supplier performance reviews
• Renewal and expiry monitoring

Each workflow follows the same operating model: Atlas completes the repetitive work, employees remain responsible for consequential decisions, and every action can be traced back to its source.

The company is now building toward a procurement operation in which routine work moves continuously, while people are brought in precisely when their authority or judgment is required.

The outcome

The manufacturer needed the work between its people, documents and systems to move without constant manual coordination.

Atlas and Foundry gave the company that operating layer, automating 80% of repetitive procurement work while leaving every purchasing decision where it belonged: with its buyers.