Here's a question worth asking the next time you scroll an app store: who actually built the thing you're about to download? Not the company. The work. Because a Business in Vancouver report published this week shows that AI agents aren't a future trend for businesses in British Columbia. They're already doing the daily grind at companies of every size, and the patterns those businesses describe are the same patterns showing up inside consumer apps right now. The report looked at three companies: Holy Duck Chili Oil, a two-person operation in Vancouver; Fantuan, a Burnaby-based Asian-cuisine delivery platform; and Campertunity, a North Vancouver service connecting campers with private landowners. Different sizes, different industries, same story. Agents do the work. Humans decide what the work should be. If you're the kind of person who reads app verdicts before tapping install, this matters to you more than it might first appear.
Key points
- A two-person chili oil company in Vancouver uses AI agents for forecasting, invoicing, and even coordinating restocking with a retailer's AI directly.
- Fantuan, a Burnaby delivery platform, runs customer-service triage, photo validation, and search improvements through AI agents.
- Campertunity's chatbot recruits and onboards campground hosts entirely on its own, no human in the loop until the deal is done.
- Fantuan's CTO argues the market now rewards generalists who can make decisions across AI-generated options, not narrow specialists.
- Holy Duck's co-founder warns that AI-generated marketing all starts to sound identical, a trust problem that hits app stores too.
Start with Holy Duck Chili Oil, because it's the most extreme example. Louise Pang and Christopher Fung run the company with, essentially, no staff. The jobs a mid-sized food company would staff with an accounting department, an operations coordinator, and a demand planner? Their AI handles them. Budgeting, invoicing, collections, reorder calls, all automated. The highest-value use case, according to Pang, is order prediction. The agent reads historical sales patterns and forecasts demand, which lets a tiny operation plan production and inventory without guessing. But the detail that should make you sit up is what happens at the retailer's end. Holy Duck's AI talks to the retailer's AI. When inventory runs low, the two agents flag it and trigger replenishment between themselves. No phone call. No purchase order typed by a human. That's machine-to-machine commerce between small trading partners, and it's live today. Translate that to consumer apps and it's not a stretch. The scheduling, resale, and delivery apps in our feed are all one step away from the same pattern: your app negotiating with a merchant's app on your behalf.
Fantuan is the clearest example of what an AI-native consumer app looks like from the inside. The delivery platform uses an agent to triage customer service, figuring out what a customer needs before a human ever sees the ticket. Courier-submitted delivery photos get validated automatically against company standards, which used to be a manual review job. The dish photography on the app gets polished by AI too. That sounds minor until you realize how much food-ordering behavior is driven by how the food looks. Fantuan's CTO, Glen Zhu, also points out that large language models have made search and recommendations noticeably better at inferring what a user actually wants, with real potential upside in conversion rates. So when you open a delivery app and the search actually understands your vague craving, that's not luck. That's the same technology stack the B.C. report describes. The line between "business AI tool" and "consumer app feature" has basically collapsed.
Campertunity took a different route. The North Vancouver company built an AI chatbot that finds prospective campground hosts, researches them, emails them, and handles the whole recruitment and onboarding conversation on its own. It also runs customer service end to end. Think about what that means for the apps you use. A support agent that resolves your issue without a human ever joining the chat is no longer a demo. It's shipping. The next time an app's support feels oddly fast and oddly fluent, there's a decent chance you're talking to software, and the honest apps will tell you so.
Zhu's take on the workforce side is the sharpest part of the report. Routine and specialized work, including coding and UI design, is increasingly something an agent can execute. What retains value is the ability to make decisions among AI-generated options: knowing which output is good, which direction serves the business, and which choice actually helps the user. He sees demand rising for generalists who can span business, product, technology, and finance, using AI to bridge those domains instead of routing work through layers of specialists. His framing is blunt: people who function only as tools for others will struggle. People who understand the business and the user, and know how to deploy AI, will do fine. If you build apps, that's a direct challenge. The indie developer who spends all week hand-polishing store screenshots is competing with someone who delegated that to an agent and spent the week talking to users instead.
Here's the counterweight, and it comes from the smallest company in the report. Pang says marketing and brand identity are the places AI can't take over. Her reason is specific: she noticed how much social content is now AI-generated, and how much of it sounds the same. Similar captions, similar phrasing, similar everything. That realization gave her a creative block, and it points at a real business risk. If your long-time audience senses a machine is speaking for the brand, they disengage. Her approach is to use AI for competitive intelligence, then deliberately craft a voice that doesn't follow the AI-generated consensus. We'd argue this applies directly to app discovery. App store listings are drowning in generated copy and near-identical screenshots. Every description promises the same seamless experience in the same confident tone. When everything sounds the same, the honest signal disappears, and users fall back on screenshots and star counts, which are easy to game. That's exactly why our verdicts here are written by people who actually opened the app, and why our vibe tags exist. A 10-second human take that says "this one's chaotic, skip it if you want calm" carries information no generated listing can fake.
Full transparency, in the spirit of the topic: Like, Apps is operated by the makers of FlirtAlly, which is currently in beta. You'll find that disclosure in our footer and about page, and you'll find FlirtAlly pinned in our grids as an Editor's app (beta) with no ratings or reviews while it's in beta. Competitor tiles get the same honest treatment, including where competitors win. If a brand's human voice is what builds trust, as Pang argues, then saying who's behind the feed is the least we can do.
The B.C. report is anecdotal, to be fair. Three companies, no hard adoption numbers, and no economist or worker perspective. But the pattern it describes is consistent with what you can see yourself the next time you open a delivery app, a booking app, or a support chat. Agents are doing the work. The apps worth your home screen are the ones honest about it, in a voice that doesn't sound like every other generated listing. Read the verdicts, check who's talking, and download accordingly.