As agentic AI moves from drafting emails to acting on a company’s behalf, firms are being told to decide in advance what it can see, do and approve
Hong Kong, October 12, 2026 – Small and medium-sized businesses in Hong Kong are being urged to set firm limits on agentic AI in business-to-business selling. These systems can pursue goals, take initiative and act across workflows, not just generate text.
For SMEs, the appeal is easy to see. Agents can run repetitive, after-hours work such as answering enquiries, chasing quotations and updating records. In B2B sales, where deals can take six months or more, they can strengthen the earliest customer touchpoints.
Guidance aimed at Hong Kong owners puts the risk in the boundaries. An agent is only as safe as the limits set on what records it can see, which actions need human approval, and where data is stored. Anything involving customer information must also comply with the Personal Data (Privacy) Ordinance. Industry commentary in the region adds that customers should be told whether they are dealing with a person or an AI.
The government is also backing adoption. It has outlined support for SMEs, including import and export firms, through funding schemes and bodies such as the Trade and Industry Department, the Hong Kong Trade Development Council and HKPC. The planned Hong Kong Artificial Intelligence Research and Development Institute is expected to offer AI security assessments and compliance consultancy.
Providers are moving in too. HKT has paired with Palo Alto Networks on security for agentic AI workloads, and Malaysia-founded Apping Tech, which builds AI agents for B2B sales, plans a Hong Kong regional hub.
