Turning Repetitive Admin into Automated Workflows with AI

Hub Distributors is a Singapore-based distribution and logistics business where a lot of the day-to-day work happens in spreadsheets, documents and inboxes. Jobs get planned and executed, products get catalogued and sent for customer review, staff get hired and onboarded, and invoices go out and payments come in.

The team that joined the programme covered four functions: Project, Warehouse, Human Resources and Accounts. Eight participants in total, each owning a different slice of the business. Project handles job admin and billing records. Warehouse prepares product photos and SKU listings for large customer orders. HR looks after recruitment, onboarding, payroll and leave admin, and compliance tracking for a high-turnover operational workforce. Accounts runs invoicing, payables and receivables, bank reconciliation and budget tracking. Putting all four in the same room was deliberate. The pain points looked different on the surface, but the root cause turned out to be the same.

Hub Distributors came in with a clear picture of the problem. Across every department, information required manual human intervention for data to be recorded into the system.

The team wanted to learn how Claude could take on that repetitive chain of work, and how to connect their spreadsheets, documents and emails into a single AI-assisted workflow instead of handling each step separately.

The repetitive work eating into the team's time included:

• Keying job scope details into Excel
• Screening applications, preparing offer letters and contracts, and tracking work pass, licence and certification renewals
• Reading, sorting and replying to a high daily volume of emails
• Typing figures from paper and PDF invoices, receipts and bank statements

The overall aim was to reduce manual data entry and duplicated work, automate selected processes where practical, and give each department a working starting point they could use straight away.

Because every department had different documents and different tools, the sessions were built around Hub Distributors' real workflows, not generic AI theory. Three principles guided the design:

Connecting Claude to real tools. Participants explored how Claude can work with Gmail, Drive and shared spreadsheets, so photo sets, invoices and HR records can be organised, matched and shared without manual copy and paste.

Building, not just prompting. Each department worked towards one working AI agent prototype, such as a project billing tracker or an invoice processing assistant, that they can keep using after the programme.

Role-Based Applications

To make sure everyone left with something relevant to their own desk, the content was tied to each function in the team:

• Project: Getting AI support on day-to-day project admin, linking job scope entry straight to the billing paperwork that follows, and cutting the double handling between spreadsheets, documents and email.
• Warehouse: Organising and naming large batches of SKU photos against the reference list, reducing mismatches, and linking spreadsheet updates directly to customer emails.
• Human Resources: Drafting job ads, JDs, HR letters and policies faster, shortlisting and summarising candidates, turning attendance, overtime and turnover data into management reports, and handling a high email volume safely and in line with PDPA.
• Accounts: Combining OCR with Claude to pull figures from paper and PDF invoices, receipts and bank statements, speeding up high-volume invoicing and bookkeeping, and moving the team from traditional bookkeeping towards a more strategic workflow.