Listen to the article
A supply-chain consultant says Claude has cut her email backlog and automated routine paperwork, freeing three to four hours a week , but only after she refined the rules to fit real-world constraints.
A Singapore supply-chain consultant has used artificial intelligence to take routine administration off her hands, saying the systems have freed three to four hours each week for client work, planning and life outside the office.
The founder of Qurious Works, who works with small and family-run businesses across South-east Asia, describes using a Claude agent to review spam and archived emails. It sorts messages by sender, checks invitations and articles against her business priorities, then recommends whether to keep a sender or unsubscribe. In its first sweep, it processed 718 messages; the weekly volume has since fallen to about 30, according to her account in e27.
The process also surfaced a weekend webinar on entrepreneurship through acquisition that she would otherwise have missed. The event led to a connection with its moderator and opened a possible new direction for her work. But the first version of the agent recommended more events than she could attend. Adding her calendar and preferred working hours helped it screen out clashes and prioritise the invitations most likely to be useful.
That adjustment reflects a broader lesson from the consultant’s experience: a tool can follow instructions accurately while still producing an impractical result if the instructions omit important constraints. Her event rules include limits on how many networking sessions to attend and how much evening time to give them, with potential clients ranked ahead of ecosystem partners and industry peers.
She has also standardised how files and AI conversations are named, making client documents easier to retrieve and reducing the risk of sending an outdated version. The e27 account says a consistent structure can also help provide relevant background for later tasks. Anthropic’s product documentation explains that Claude Projects can draw on material deliberately added to a project, although it does not guarantee that information will always be retrieved or interpreted correctly.
A similar effort has been applied to a food-and-beverage client’s paperwork. Supplier invoices arriving as photographs in WhatsApp groups are now downloaded, renamed according to a common format and placed in a shared folder for the accounts team. A separate tool compares the prices on invoices with recorded prices, helping identify changes that have not been announced. The example illustrates how AI-assisted workflows can address repetitive handling as well as information retrieval.
For creative work, the consultant uses AI differently. Rather than asking for one workshop plan, she requests several formats and asks the model to estimate how likely each is, with some options required to be less predictable. The approach draws on Verbalised Sampling, a research method that asks language models for multiple responses and associated probabilities. The researchers found greater variety in creative-writing tests, but their findings do not validate the probability estimates or the workshop-planning process described by the consultant.
The consultant says AI supplies starting points, while she chooses and combines exercises and adapts them to the client’s team. That division of labour is consistent with a separate study of creative metaphor writing, which examined how people and AI assistance affect collective variety. The study did not test business workshops, so it offers context rather than direct evidence for this particular use.
The account’s productivity case sits alongside a wider challenge for smaller firms. Singapore’s Ministry of Manpower reported that 23.9 per cent of firms with fewer than 25 employees had adopted AI, compared with 76.4 per cent of firms employing more than 500 people. Across firms of all sizes, the reported adoption rate was 28.5 per cent, with uptake also varying by sector.
The consultant says she now divides the time recovered among taking on more work, focused thinking, family and tennis. Her advice to other small businesses is to begin with an unwanted task, define the rules before automating it and revise them when real-world use exposes gaps.
Source Reference Map
Inspired by headline at: [1]
Sources by paragraph:
Source: Fuse Wire Services


