
If you run a small business, you've probably had the conversation by now. Someone tells you AI agents are the future, you nod, and then you wonder what one would actually do all day in your company. A report published this week by Business in Vancouver offers a useful answer, because it skips the theory and shows what agents are already doing inside ordinary B.C. businesses, including one with exactly two people on staff.
Key points
- A two-person Vancouver condiment maker uses AI agents for budgeting, admin, and order forecasting, work that larger firms staff with whole departments.
- One small business reports its agent communicating directly with a retailer's agent to manage inventory and trigger reorders.
- Fantuan, a Burnaby delivery platform, uses agents for support triage, courier photo validation, and repetitive internal tasks like meeting summaries and translation.
- Campertunity's AI chatbot handles prospect research, outreach emails, and onboarding for campground hosts.
- Employers say hiring is shifting toward generalists who can direct AI toward business goals, while owners keep brand voice and judgment in human hands.
Holy Duck Chili Oil is a family condiment business in Vancouver run by co-founders Louise Pang and Christopher Fung. According to the report, published October 22 and picked up by several B.C. community outlets, the company uses AI for budgeting, administration, and order forecasting. The agent analyzes past ordering patterns to predict future demand, which feeds directly into production and inventory planning. Pang described the predictions as generally accurate. Think about what that means structurally. Forecasting and operations planning are roles a large corporation would assign to dedicated accounting and operations teams. A two-person company can't hire those teams, and until recently it simply absorbed the work or went without the insight. An agent changes that math. The interesting part isn't the technology itself, it's who gets to use it. Functions that used to require headcount are now available to owner-operators, and the businesses adopting them describe outcomes, not build processes. None of the companies profiled talk about assembling agents themselves. They talk about what got handled.
The most quietly significant detail in the report comes from Pang's description of how Holy Duck works with a retail partner. Her AI agent and the retailer's agent communicate inventory levels between the two businesses and trigger reorders automatically. No phone call, no email chain, no spreadsheet reconciliation. This is an early, concrete example of agent-to-agent commerce, and it's worth sitting with. If your suppliers, distributors, or retail partners start running their operations through connected agents, the question shifts from whether an agent would help your business to whether your business can participate in that kind of exchange at all. A company whose inventory data lives in someone's head or in an unstructured inbox can't join that conversation. A company with a connected agent can. We'd frame this as a trend observation rather than a prediction. But it points at something real: over time, having an agent that can talk to other systems and other agents starts to look less like an efficiency choice and more like a prerequisite for certain kinds of business relationships.
The mid-size example in the report is Fantuan, the Burnaby-based Asian-cuisine food delivery platform. CTO Glen Zhu described a familiar division of labor. AI handles customer service triage, sorting inquiries before a human gets involved. It automates dish-photo enhancement on the app, work that used to be labor-intensive. An agent validates courier delivery photos against company standards. Internally, agents handle the repetitive layer: organizing information, meeting summaries, translation, content drafting, and data analysis. That list will look familiar to any small business owner, because it's the same list of tasks that eats an afternoon. Triage the inbox. Clean up the photos. Check the submissions. Summarize the meeting. None of it is hard, all of it is constant, and every hour spent on it is an hour not spent on customers or growth. The sales side shows up in the Campertunity example. The North Vancouver platform, which connects campers with private landowners, uses an AI chatbot to research prospective campground hosts, send outreach emails, and handle recruitment, onboarding, and customer inquiries. That's prospecting and follow-up running in the background, with the human team stepping in for judgment calls. Zhu was careful to frame outcomes like improved search relevance as potential rather than guaranteed, which is the right instinct and one we share.
The report's central workforce argument comes from Zhu, and it's blunter than most corporate commentary. He argues that demand is moving toward generalist employees, people who understand the business, the users, the technology, and the finances, and who can use AI to bridge gaps between those areas. Specialists who simply execute assigned tasks are the ones whose value thins out. In his words, people who function purely as tools for others will struggle. He also suggests team structures may flatten. Instead of a request passing through layered specialist teams, a small, broadly skilled group could carry a project end to end with agents filling the gaps. For a five-person company that was never going to hire a specialist for anything, this is mostly good news. The skill that matters is the one owner-operators already have: knowing your business well enough to say which parts of it an agent should handle, and which parts need a human. AI generates the options. You still decide. That division doesn't shrink your role, it changes what you spend your time on.
The report's most useful counterpoint comes from Pang, who deliberately keeps marketing and brand identity out of AI's hands. Her reason is specific: AI-generated social content tends to look and feel the same across everyone who uses it, and she doesn't want her brand to blend into that. She'll use AI to see what everyone else is doing, then intentionally craft a distinct voice, because audience trust depends on it. We think this is the right instinct, and it's worth being explicit about why. Agents are excellent at the routine and the repeatable: answering the phone, following up on leads, summarizing, organizing, forecasting. They're the wrong tool for deciding what your brand sounds like or what your business stands for. The healthiest setup we see keeps a clear line between the two. The agent handles the operational floor so the owner can spend real attention on the things only the owner can do. That's also why ownership matters. When an agent answers your phones or writes your follow-up emails, it's representing you. Businesses that keep control of their agents, their data, and their voice, on terms they can walk away from, are the ones that get the upside without the risk Pang is describing.
The B.C. report is worth reading less for its technology claims and more for its casting. The businesses using agents well are a two-person condiment maker, a delivery platform, and a camping marketplace. They're ordinary companies doing ordinary work, and the agents are handling the ordinary parts: forecasting, triage, follow-up, onboarding, reordering. Meanwhile the humans are doing the things the report says humans should keep: strategy, judgment, and brand. If you've been waiting for a signal about whether this applies to a business your size, that's the signal. The question isn't whether agents can work in a small company. It's which three or four recurring tasks you'd hand off first, and who you'd trust to build and run them.