Port and Aquarium.

The US Census American Community Survey for 2024 reports that 42.8 percent of Long Beach residents speak a language other than English at home. The breakdown is unusual for a coastal California city of LB's size: roughly twenty-eight percent of LB residents are Hispanic or Latino, primarily Mexican-American with a smaller Salvadoran and Guatemalan community; about thirteen percent of LB is Asian, with the Cambodian-American community concentrated in central LB and a sizable Filipino-American community spread across Bixby Knolls, the east side, and the Carson border. Vietnamese-American communities cluster in west LB and along the Lakewood border.

The language distribution at the phone-call level is sharper than the population numbers suggest, because language-of-ordering correlates with the cuisine and the neighborhood, not just with the speaker's general fluency. A Mexican-American family ordering from a pupuseria on Anaheim Street will place the order in Spanish even if everyone in the household speaks fluent English at work. A Filipino-American family ordering from a Cebuano restaurant in Bixby Knolls may place the order in Tagalog or Visayan. A Cambodian-American family ordering kuy teav from Phnom Penh Noodle Shack places the order in Khmer. The language of the food becomes the language of the call.

The operational implication is non-trivial: a Long Beach restaurant phone line that only handles English drops twenty to forty percent of inbound calls depending on the corridor. A line that handles English and Spanish drops less, maybe ten to twenty percent, but still misses the Khmer-speaking, Tagalog-speaking, Vietnamese-speaking, and Cantonese-speaking callers. The full multilingual Voice AI surface is the only configuration that captures the inbound. Our LB pilot operators report a 31 to 37 percentage point lift in phone-to-order conversion when the Voice AI handles all five languages (Khmer, Spanish, Tagalog, Vietnamese, English) versus an English-only IVR.

There is also a community-cultural dimension we have learned to design for. The Khmer-speaking Cambodian-American community in Long Beach skews older for first-generation and primary-language speakers (the genocide-era refugees are now in their sixties and seventies); they are less likely to use a mobile app for ordering and more likely to place a phone call. The Spanish-speaking community in LB has a wide age distribution and is comfortable with both apps and phone orders. The Filipino-American community is the highest mobile-app penetration of the four groups. The right product configuration is not "everyone uses the app"; it is "the app is available for those who want it, the phone is available for those who need it, and both lead to the same order in the same kitchen ticket."

The Cambodia Town Voice AI requirement becomes a captured-order economy when Khmer, Spanish, Tagalog, Vietnamese, and English are all live on the same restaurant phone number. The Phnom Penh Noodle Shack operator no longer drops 51 percent of her inbound calls to voicemail. The cook on the line still cooks; the multilingual Voice AI handles the order intake, confirms the menu items in the caller's language, and writes the order to the same kitchen printer. The conversion lift in our pilot was 31 to 37 percentage points. The technology is not optional for these operators; it is the platform.

The Port economy becomes a daytime catering channel when an LB restaurant can accept net-30 invoiced orders from port-administrative offices, ILWU dispatch, customs brokers, and shipping line client lunches. The marketplace apps do not handle invoicing, secure-perimeter delivery, or the deposit-paid catering tickets the port runs on. DirectOrders does. The five or six restaurants serving the port catering economy run $40,000 to $120,000 monthly in catering alone; the platform pays for itself in a single Tuesday.

The Grand Prix weekend becomes a five-day pre-order revenue event when the restaurant has a working catering form, a pickup-window scheduling tool, and the ability to push a "Grand Prix weekend hours" banner to customers. The restaurants that handle the Grand Prix well book the bulk of their April revenue across those five days. The restaurants without a direct ordering platform lose the week entirely or worse, pay marketplace commission on the catering tickets that should have been booked direct.

The Aquarium and Queen Mary tourist flow becomes captured pickup and delivery demand when restaurants rank for the corridor phrases ("lunch near Aquarium of the Pacific", "Pine Avenue Long Beach dinner", "Belmont Shore brunch"). A branded ordering page indexed independently of the marketplace aggregator pages, with menu photography and clear hours, beats the aggregator slots on most non-paid queries. The flat $249 monthly cost is repaid in two captured tickets per week from search-driven traffic.

And the LA-OC border position becomes a strategic advantage rather than a confusing geography when the restaurant's ordering page is geography-agnostic. The customer in Huntington Beach who searches "Long Beach catering 2nd Street" finds the restaurant's direct page; the customer in Lakewood searching "Bixby Knolls breakfast pickup" finds the same. The marketplace apps cannot do this because their promotional zones are bounded; the direct page does it natively. Long Beach restaurants positioned for both the LA County and OC County customer base capture twice the addressable market when the platform supports it.