Singapore Global International Adopts AI Agents to Explore a Lean “One-Person Company” Model

How a small team runs four business lines

“Can one person plus a set of AI agents keep a multinational running day to day?” That is the operating experiment Singapore Global International has recently started internally. The idea is simple: hand repetitive, process-driven work to AI agents, and keep people on the parts that need judgement and accountability. 1. Why learn from the “one-person company” A “one-person company” (OPC) does not mean a company with only one employee. It describes an operating style built on small teams, high automation and low fixed cost — replacing headcount with tools and systems so the organisation stays light and fast. For a company running four business lines — International Education, International Trade, AgriTech and Smart Dining — that thinking is especially relevant: each line has its own rhythm, but much of the back-office work is shared. 2. Four pilot scenarios today International Education: common enquiries about courses, visas, accommodation and fees are turned into a standard answer base; agents handle the first response and advisers take only the cases that need individual judgement. International Trade: enquiries are classified automatically, key details extracted (category, quantity, destination port, lead time) and a draft quotation prepared for the sales team to confirm before it goes out. AgriTech: project files, growing parameters and supplier information are archived in one place and retrieved on request, instead of being searched for again and again. Smart Dining: equipment and order data from stores are summarised into a daily report, with anomalies flagged so operators can focus on what genuinely needs a human. 3. What it means for clients Faster first responses, a more consistent message, and human time spent where professional judgement actually matters. Every external commitment and quotation is still confirmed by a person before it is sent. 4. What comes next The company plans to extend agent workflows to contract and document version control, and to formalise an internal “human review + machine execution” standard: agents take the repetitive work, people make the final call.