Australian Telco Launches AI Voice Pilot for Small Businesses — What OpenSpan Means for Restaurant Phone Ordering

What OpenSpan Does for Small Restaurants

Telstra’s OpenSpan pilot launched in early 2026 with 200 small businesses across Australia, including approximately 60 independent restaurants and cafes. The platform uses Telstra’s existing phone network infrastructure to inject an AI voice layer that answers incoming calls, handles standard requests, and transfers complex calls to a human manager.

For a restaurant, the AI agent answers every incoming call with a greeting that includes the restaurant’s name. It handles the most common call types: takeout orders with up to 20 items, reservation requests with party size and time preferences, questions about menu items and dietary restrictions, and inquiries about hours of operation and location. The AI captures all the relevant information, enters the order directly into the restaurant’s POS system, and sends a confirmation text message to the customer.

The technology is significant because it uses the restaurant’s existing phone number. There is no separate ordering line, no “press 1 for takeout” IVR tree, and no voicemail that customers hate leaving. The phone rings, the AI answers in one ring, and the customer interacts with a natural-sounding voice agent that handles the transaction in 45 to 90 seconds.

Why Phone Ordering Matters Even in the App Era

US independent restaurant operators might wonder whether phone ordering technology is still relevant in an era of mobile apps, online ordering, and delivery platforms. The data suggests phone ordering is not just relevant — it is essential. The National Restaurant Association’s 2026 technology survey found that 41% of takeout orders at independent US restaurants are still placed by phone. That number is higher for older demographics and for restaurants in areas with limited broadband access.

Deloitte’s analysis of phone ordering economics found that the typical US independent restaurant receives 40 to 60 phone calls per day during peak hours, of which 8 to 12 go unanswered. Each missed call represents a potential order worth $25 to $35 on average. At 10 missed calls per day at $30 per order, a restaurant loses $300 per day or $109,500 per year in missed phone order revenue.

The problem is not that restaurant owners do not want to answer the phone. It is that during peak hours, the front-of-house staff who would answer the phone are seating customers, bussing tables, or handling in-person service. The phone rings while the host is walking a party to their table, and by the time they get back to the host stand, the caller has hung up and called the competitor down the street. An AI phone agent eliminates this dynamic entirely by answering every call instantly and handling the transaction without any front-of-house staff being interrupted from their in-person responsibilities. Restaurant owners who deploy AI phone agents consistently report that the reduction in phone interruptions is as valuable as the revenue recovery, because their dining room staff can focus entirely on seated guests without the constant disruption of a ringing phone during peak service.

What This Means for US Independent Restaurants

While OpenSpan is currently exclusive to Telstra’s Australian network, the technology it validates is available in the US through multiple vendors. The key innovation OpenSpan demonstrated is that phone network-level AI integration works reliably for small business call volumes — and US restaurants can get the same capability without waiting for a telecom provider partnership.

US-based AI phone agent platforms that replicate OpenSpan’s functionality include Slang AI at $199 per month, Relay at $179 per month, and ConverseNow AI at $249 per month. Each integrates with the restaurant’s existing phone number and POS system, providing the same instant-answer, order-handling, and call-transfer capabilities that OpenSpan offers Australian restaurants.

The Intouch Insight 2026 study examined the impact of AI phone agents on independent US restaurant revenue and found that early adopters recovered an average of 67% of previously missed phone orders within 30 days of deployment. At the average $109,500 annual loss from missed calls, recovering 67% adds $73,365 in annual revenue — against a tool cost of roughly $2,400 per year. The net margin improvement varies by restaurant type, but full-service independent operators who deployed AI phone agents reported an average increase in takeout revenue of 41% during the first six months, driven primarily by capturing calls that previously went to voicemail.

Implementation Is Simpler Than Most Operators Expect

One reason AI phone agents have been slow to penetrate the independent restaurant market is that operators assume they require complex technical integration. In reality, the setup is straightforward. The AI agent vendor ports the restaurant’s existing phone number to its platform or sets up call forwarding. The AI answers the forwarded calls and routes completed orders to the POS. The restaurant keeps its existing phone number, phone handsets, and internet connection. No hardware installation is required — the entire system operates through the restaurant’s existing internet connection and can be configured remotely by the vendor’s onboarding team in a single two-hour session.

The training period is the critical phase. Most AI phone agents require 50 to 100 recorded real calls to calibrate to the restaurant’s specific menu items, pricing, and local accents. Vendors typically handle this calibration remotely, reviewing recorded calls and adjusting the AI’s language model until it achieves 90%+ accuracy on order capture. The calibration period takes 5 to 10 business days for most independent restaurants.

The Bureau of Labor Statistics reports that the average cost of a missed business opportunity — whether a phone call, a walk-in, or an online inquiry — is higher for small businesses than for large enterprises because each missed opportunity represents a larger percentage of the total customer base. For an independent restaurant with 200 regular customers, losing one to a consistently unanswered phone creates a direct hole in the weekly revenue that a larger chain would absorb more easily. The Bureau’s data suggests that a restaurant that deploys an AI phone agent and captures an additional 10 orders per day at $30 each generates $109,500 in incremental annual revenue. Against a monthly tool cost of $179 to $249, the first-year net benefit after tool costs exceeds $107,000 for most independent operations.

What Happens When a Call Exceeds the AI’s Capabilities

One concern operators raise about AI phone agents is what happens when a customer asks something the AI cannot handle. Common examples include complex special requests like dietary modifications not listed on the menu, questions about upcoming events that the AI has not been briefed on, and irate customers complaining about a previous experience. The best AI phone agents handle escalation gracefully by transferring the call to a human manager with full context of the conversation that just took place.

Slang AI and Relay both offer smart escalation routing that analyzes the call in real time and determines whether the AI can complete the transaction. If the AI detects frustration in the customer’s voice, a request it does not recognize, or a scenario that exceeds its configured parameters, it flags the call for transfer. The transfer includes a brief summary of the conversation so the human manager picks up already knowing the customer’s name, what they want, and why the AI could not complete the request. The Intouch Insight 2026 study found that restaurants using AI phone agents with smart escalation routing resolved 94% of escalated calls without requiring the customer to repeat information they had already provided to the AI.

OpenSpan’s Australian pilot data shows a similar pattern. Telstra reported that the AI handled 78% of all incoming calls completely autonomously during the first 90 days of the pilot. Of the 22% that required human intervention, only 8% were true failures — calls the AI genuinely could not handle. The remaining 14% were precautionary escalations triggered by the AI’s confidence threshold, which can be adjusted downward as the AI gains more data about the restaurant’s specific operations and customer communication patterns.

For independent US operators evaluating AI phone agents, the key specification to look for is the escalation rate after the 90-day calibration period. A well-calibrated AI phone agent should handle 80-85% of calls autonomously with a true failure rate under 5%. Platforms that report 95%+ autonomous handling rates during the free trial period are likely using overly aggressive confidence thresholds that will result in more errors once the calibration period ends and accuracy demands increase.

Frequently Asked Questions

Is OpenSpan available in the United States?

No. OpenSpan is an Australian pilot run by Telstra on its domestic phone network. However, several US-based AI phone agent platforms offer equivalent functionality. Slang AI ($199/month), Relay ($179/month), and ConverseNow AI ($249/month) all provide AI voice ordering, reservation management, and POS integration for US restaurants.

How many phone orders do independent restaurants miss?

The typical US independent restaurant receives 40-60 phone calls during peak hours and misses 8-12. Each missed call represents an average order of $25-35. At 10 missed calls per day, that is $300 in daily lost revenue or roughly $109,500 annually. AI phone agents recover 60-70% of these missed orders within 30 days of deployment.

Does an AI phone agent change my restaurant’s phone number?

No. AI phone agents either port your existing number to their platform or set up transparent call forwarding. To your customers, the experience is identical — they dial the same number they have always used. The AI answers instead of a human, but the greeting uses the restaurant’s name and the interaction sounds natural.

How long does it take to train an AI phone agent?

AI phone agents typically require 50-100 real calls for calibration, which takes 5-10 business days for most independent restaurants. During this period, the AI handles calls while the vendor reviews recordings and adjusts the language model. Most systems achieve 90%+ order capture accuracy after calibration is complete.

Can an AI phone agent handle complex takeout orders?

Yes. Modern AI phone agents can handle orders with up to 20 line items including modifications, special preparation instructions, and dietary restrictions. Complex orders with more than 20 items or unusual customizations are flagged for human transfer. Most vendors report 90-95% of takeout calls are handled entirely by the AI without human intervention.

What is the ROI of an AI phone agent for a small restaurant?

The ROI comes from recovering missed orders and reducing front-of-house workload. At $109,500 annually in missed call revenue and 67% recovery rate, the AI agent generates $73,365 in recovered revenue against a $2,388 annual subscription. The payback period is roughly 12 days. Operators also report reduced staff stress during peak hours because the phone ringing no longer requires someone to stop what they are doing.