On a vibrant evening in Melbourne, located at Crimson Coffee on the iconic Camberwell main street, the inaugural Founders MeetUp, hosted by Jacob Cheng, Founder of Trafficon and a leading Chinese-Australian Digital Specialist, brought together a diverse group of entrepreneurs to discuss innovation and opportunity.
The guest speaker, Yifei Wang, R&D expert, lecturer, and founder of AIBUILD, provided attendees with insights on leveraging AI for business growth. The event featured professionals from various industries—finance, real estate, hospitality, and more—each sharing unique perspectives. Below is an edited transcript of the engaging conversation between Jacob (Q) and Yifei (A), with contributions from attendees.
Q: Yifei, many educators and professionals talk about AI. What opportunities have you seen in the market?
A: Thank you, Jacob, and it’s great to hear from everyone here. As someone with 15 years in education, translation, and now AI development, I’ve seen AI tackle business bottlenecks, but it’s not as “smart” as we often assume. Current tools—like AI-powered websites—improve efficiency but miss the core potential: large-scale data utilization. Industries like cybersecurity, tech, and even those represented here today can use data to address competitors’ pain points. The real opportunity lies in building databases to drive innovation—like selling shovels to gold miners.
Q: Any case studies to illustrate this?
A: Absolutely. In education, where I teach over 100 students at a time, AI helps me track progress in ways I couldn’t manually. For instance, we’ve developed systems where students generate barcodes to access assignments in virtual environments. This allows me to analyse their thinking and provide tailored feedback.
Another example is KT Mart, a Melbourne-based chain with three CBD stores. They had no online platform, so we built an app that now has 40,000 users—nearly half their active market. AI analyses stock history to recommend order quantities, reducing warehouse costs. For supermarkets, faster sales mean higher profits, but emotional decisions often lead to waste. Our system considers background data—like new product launches and customer interest—to optimise orders. This isn’t just internal; KT Mart’s model can be sold to other retailers as a solution. Every industry is unique, but the key is solving your niche problem and commercialising it.
Allan (Financial Trading): I agree—data is critical. KT Mart’s example is compelling, but can this apply to hospitality? My clients struggle with turnover in remote retail shops.
A: Great question, Allan. Many supermarkets, like couple-run businesses, aim to cut costs. We’re matching visual models with existing databases to provide accurate product insights without manual data entry. Some owners hide wholesale prices from staff, so we use sensors and POS systems to gauge supply and demand. For hospitality, similar principles apply—AI can streamline stocktaking and ordering, but it’s about integrating with your specific database.
Q: AI agents are a buzzword. Can you elaborate on their role?
Kevin (Real Estate): I’ve seen landlords use AI agents for property inspections, eliminating the need for human agents to schedule viewings.
A: Exactly, Kevin. Think of an AI agent as an apprentice. A property agent passes knowledge—like scripts—onto the AI, which learns continuously. It may not reach expert-level intuition, but with enough data, it gets close. The agent’s potential depends on market size and industry complexity. If there’s demand, there’s a case for developing it. Like an apprentice becoming a master, the AI can scale to create “avatars” for repetitive tasks, but it needs clear instructions—not just prompts, but coding.
Q: So, coding over prompts?
A: Yes. Prompts are the last step. Effective AI requires tiered coding and data analysis to filter outliers. It’s not about “big data” hype but building systems tailored to your needs. This requires a team—product managers, developers, and your expertise to translate market pain points.
Jane (Entrepreneurial Interviews): AI can’t yet replicate advanced expertise, but once you build an agent, can it multiply itself beyond human limits?
A: Precisely, Jane. The initial effort is huge, but once built, the cost per “avatar” drops. In Australia, labor costs are high, so developing an agent might start at $50,000, ranging up to $500,000 depending on complexity. Australia’s innovation ecosystem—fuelled by government grants and mining revenue—supports this. Unlike Singapore or East Timor, our government invests directly in applied research, which is rare.
Peter (CBA Homeloans): I work in homeloan applications, where the human touch is key. How long before AI replaces me?
A: Peter, AI won’t replace tailored services anytime soon. It can perform risk analysis to support decisions—like loan approvals—but it doesn’t take legal responsibility. Doctors, for instance, still make final calls despite AI diagnostics. In Chinese, “danger” and “opportunity” are the same word. AI is a tool to enhance, not replace, your expertise.
John (Shopping Centre Business): If I invest in an AI agent, what about IP protections?
A: Good question, John. I don’t recommend everyone builds from scratch. Existing tools might suffice, so evaluate before disclosing sensitive data to competitors. My ambition is an Australian OpenAI, keeping our data local for security. For smaller businesses, focus on niche needs—solve your problem, then scale the solution. Canva, for example, guards its data tightly. If your solution is profitable and marketable, IP protection becomes worthwhile.
Ben (Hospitality): In 2016, I explored digital electricity control systems for restaurants. Stocktaking, rostering, and ordering are pain points. Could AI screen supplier data or create a digital shop assistant? What’s the cost?
A: Ben, you’re spot-on. AI can optimize stock—like calculating losses on 1kg of pork for sweet-and-sour dishes. Smart meters already exist for group buying to negotiate better rates. For marketing, imagine a POS system offering credits for loyalty—say, 20% off if you spend $60. No current POS does this well, but solving it could be sold to others. Costs vary, but it’s about linking your expertise to AI’s scalability.
Jane: This reminds me of AMEX’s credit offers. You started in AI in 2017—what a time!
A: Yes, 2017 was ideal—AI was powerful but not overhyped. I filed patents in computer vision, though they’re now easily replicable. Australia’s funding model supports innovation at the application level, unlike China or the US, which focus on basic research. The revolution is coming, but general AI matching a 20-year expert is still far off.
Q: AI often struggles to verify its accuracy. Thoughts?
Allan: My friends in quantum computing say AI opens opportunities, but verification requires multiple experts. Basic jobs may be replaced, but new roles emerge.
A: Exactly, Allan. Second opinions matter. AI can handle basic tasks, but complex solutions need human expertise to verify and coordinate. It’s about collaboration, not replacement.
We Look Forward to Seeing You at Our Next MeetUp
As Yifei aptly put it, “Solve your problem, then sell the solution.” With Australia’s unique innovation ecosystem, the stage is set for founders to turn pain points into profits. Thank you for joining us at our first inaugural Founders MeetUp event, and stay tuned for what we have next!
