How Bespoke faked AI until it actually worked (w/ Akemi Tsunagawa) | E2320

Episode Summary

Executive Summary: Akimi Soagola discusses how Bespoke evolved from a multilingual tourism chatbot into a broader labor-tech platform addressing Japan’s twin challenges: surging tourism and a shrinking workforce. She explains how her company trains foreign workers, automates inspections with robots, and uses AI tools internally, while also highlighting Japan’s cultural and regulatory tensions around immigration, language, and overtourism.

Main Topics: Japan’s tourism boom and overtourism (Priority: 5/5): The conversation opens with Japan’s rapidly growing tourism market, its drivers, and the friction it creates in hotels, restaurants, neighborhoods, and transport spaces. Bespoke’s origin in multilingual chatbot technology (Priority: 5/5): Akimi explains how her first company solved traveler communication problems in Japan using a chatbot built in-house before generative AI became mainstream. Labor shortages and foreign worker onboarding (Priority: 5/5): The discussion shifts to Japan’s demographic decline, labor shortages, and how Be Trained helps companies train non-Japanese workers with smart glasses, video, and quizzes. Robotics for industrial inspections (Priority: 4/5): Akimi describes a third business line focused on automating scaffold and shipyard inspections with robots to reduce accidents and replace repetitive manual labor. Cultural integration and language requirements (Priority: 4/5): The episode examines the tension between preserving Japanese social norms and integrating foreign workers, including language-test requirements and local government concerns. Bootstrapped growth, profitability, and bank financing (Priority: 4/5): Akimi explains how the company raised limited venture capital, became profitable, and later leveraged government contracts to secure bank loans. AI as an internal operating system (Priority: 3/5): The conversation closes with how Akimi’s company uses AI heavily for coding, job descriptions, summaries, and staffing decisions, reflecting a broader startup productivity shift.

Key Arguments: Japan’s weak yen, high quality, food, shopping, and culture have made it much more attractive to tourists than it was in 2015. Overtourism is not just a current issue; it has intensified since 2017-2018 and is causing real local tension and rule enforcement challenges. Foreign workers are becoming essential for Japanese industries such as construction, shipyards, hotels, restaurants, and nursing. Language and training barriers make onboarding foreign labor difficult, so companies need better tooling for vocational education and compliance. Many safety and inspection tasks in industrial settings are repetitive and should be automated with robotics rather than left to overworked human teams. Japan’s population decline is creating not only labor shortages but also business succession and regional hollowing-out problems. AI tools now let the company do with fewer people what previously required new hires, changing how startup teams should think about headcount. Government contracts can serve as collateral for bank loans in Japan, making non-dilutive financing possible once a company is profitable.

Data Points: Tourist arrivals in Japan: 60 million per year - Akimi cites current tourism scale when discussing how much demand has grown. Founder University cohort count: Second cohort completed; third cohort planned for January - Jason mentions the program’s expansion in Japan. Angel University participants: Close to 20 angel investors - Describes the Tokyo angel investor gathering. Population loss: Equivalent to Yamanashi Prefecture each year; around 700,000 to 800,000 people, rising toward 1 million - Used to illustrate Japan’s demographic decline. Company headcount: 65 people - Akimi notes that more than half are non-Japanese. Non-Japanese employees: More than half of 65 employees - Highlights the company’s international workforce. Initial venture funding: 450 million Japanese yen - Funding raised for the original chatbot business. Approximate USD equivalent of funding: About $3 million - Host translates the 450 million yen raise into U.S. dollars. Loan term: Up to 10 years - Bank financing secured against government contracts. Loan interest rate: Around 1% - Akimi describes favorable financing once profitable. Foreign language test requirement: N4 level - Discussed as the minimum language requirement for some incoming workers. Shipyard output decline: From 9 ships per year to 6 ships per year - Due to labor shortages, production capacity has fallen. Paint inspection labor: More than 1,000 human hours - Example of a repetitive industrial inspection task that could be automated.

Pivotal Quotes: "We did a lot of test run, right? We also had humans, including myself, chat with people pretending to be a bot." — Akimi Soagola: Explaining how the company validated demand before chatbot tools were widely accepted. "So, one of the things that we looked into first was scaffolds. So, scaffolds are everywhere at construction site, shipyard, and scary." — Akimi Soagola: Introducing the robotics use case focused on industrial safety inspections. "So, have you thought about expansion plans globally, or you have so much work here to do?" — Host: Prompting discussion of whether the business model can scale beyond Japan.

Implications: The episode shows how demographic decline, tourism growth, and AI/robotics adoption are converging into a major Japan startup opportunity. For founders, the lesson is to build from real operational pain points and use automation to offset labor scarcity.