Automating Manual Workflows with Claude

Katrin BJ Pte Ltd is a Singapore-based company with a lean, multi-functional team spanning management, accounts, administration and sales. Like many established SMEs, the business runs on a tight set of day-to-day operations: preparing quotations for customers, issuing invoices, managing stock, and keeping records accurate and up to date.

The team that took part in the programme reflects how the business actually runs. It brought together the Director and General Manager, who steer operations and make the final call on process changes, alongside the Account Manager and Accounts Executive handling billing and financial records, Admin Executives and an Admin Assistant keeping the back office moving, and two Sales Executives who sit on the front line with customers.

Having every function in one room was a deliberate choice, since the biggest efficiency gains usually sit in the handoffs between roles, not inside any single one.

Katrin BJ came to Hustle with a clear and honest starting point: roughly 90% of their operational workflows were still manual, with heavy reliance on human input at almost every step of the business.

The team had some exposure to AI, mainly using ChatGPT and Gemini for simple, one-off tasks. But AI had not yet made its way into the actual processes that run the business. There were no automated workflows, no internal tools built around their operations, and no clear picture of what was possible beyond typing a question into a chat box.

The goal was to change that. Katrin BJ wanted to understand how Claude could be applied well beyond basic prompting, and how it could help the team build practical internal solutions such as SaaS-style dashboards, workflow automations and streamlined operational processes.

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

• Preparing and generating invoices
• Preparing and issuing quotations
• Stock-taking and inventory-related processes
• Administrative and operational data handling
• Other repetitive day-to-day business tasks

The overall aim was to reduce repetitive manual work, improve process efficiency, minimise unnecessary human intervention, and lift overall productivity through AI that people could actually use on Monday morning, not in theory.

Because the team's needs were so tied to real operations, the training was shaped around practical workplace use rather than generic AI theory. Three principles guided the design:

Beyond basic prompting. Most of the team already knew how to ask an AI a question. The focus here moved to the next level: giving Claude proper context, working with real business documents, structuring repeatable prompts, and turning one-off answers into reliable processes.

Workflow transformation, not just tool training. Each manual process was treated as a workflow to be redesigned. Participants looked at how a quotation or invoice currently moves from request to final document, where the time goes, and where Claude can take over the repetitive parts while people keep control of the decisions that matter.

Building, not just using. A key part of the session explored how Claude can help non-developers create simple SaaS-style dashboards and internal tools, so tasks like tracking quotations, monitoring stock levels or generating billing documents could eventually run through a purpose-built interface rather than scattered spreadsheets and emails.

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:

• Management (Director and General Manager): Using Claude to summarise business data, draft reports, spot process bottlenecks, and scope out internal tools worth building first.
• Accounts (Account Manager and Accounts Executive): Streamlining invoice preparation, checking figures and records for consistency, and reducing the back and forth involved in billing and documentation.
• Admin (Admin Executive and Admin Assistant): Handling repetitive data entry, document preparation, filing and operational data tasks faster and with fewer errors, and supporting stock-taking and inventory workflows.
• Sales (Sales Executives): Speeding up quotation preparation, drafting customer emails and follow-ups, and keeping sales information organised so quotes go out faster and more consistently.