By Sagar Shankaran, Founder of CallSphere
How restaurants use AI voice agents to handle phone reservations, eliminate OpenTable fees of $1-7.50/cover, and own their customer data.
Key takeaways
Every restaurant owner knows the math, even if they try not to think about it. OpenTable charges $1.00 per network cover (guest books through OpenTable's website/app) and up to $7.50 per cover for premium placement. Resy charges restaurants a flat monthly fee of $249-$899 depending on the tier, plus transaction fees. Yelp Reservations, Google Reserve, and similar platforms each take their cut.
For a 120-seat restaurant doing 2 turns per night, 6 nights a week — roughly 1,440 covers per week — the OpenTable bill alone ranges from $1,440 to $10,800 per week, or $75,000 to $561,600 per year. Even at the lower end, this is a massive line item for an industry that operates on 3-9% net profit margins.
But the cost extends beyond fees. When a guest books through OpenTable, OpenTable owns that relationship. They market competing restaurants to your guests. They control the review narrative. And they can change their pricing at any time, because switching costs are high once your guest database lives on their platform.
The alternative has always existed: answer the phone and take reservations directly. The problem is that restaurants cannot answer the phone. During service — which is exactly when most people call to make reservations — every staff member is occupied with guests in the room. Industry data shows that 62% of restaurant phone calls go unanswered during peak hours (5-9 PM). Those missed calls drive guests to OpenTable, which answers the phone with a booking page.
AI voice agents break this cycle. They answer every call, take reservations 24/7, and the restaurant keeps 100% of the customer data and pays zero per-cover fees.
Restaurant operators understand the fee structure is unfavorable. Yet switching away from OpenTable and Resy is rare. The reasons form a self-reinforcing loop:
Hear it before you finish reading
Talk to a live CallSphere AI voice agent for restaurant in your browser — 60 seconds, no signup.
flowchart LR
CALLER(["Diner"])
subgraph TEL["Telephony"]
SIP["Twilio SIP and PSTN"]
end
subgraph BRAIN["Restaurant AI Agent"]
STT["Streaming STT<br/>Deepgram or Whisper"]
NLU{"Intent and<br/>Entity Extraction"}
TOOLS["Tool Calls"]
TTS["Streaming TTS<br/>ElevenLabs or Rime"]
end
subgraph DATA["Live Data Plane"]
CRM[("CRM and Notes")]
CAL[("Calendar and<br/>Schedule")]
KB[("Knowledge Base<br/>and Policies")]
end
subgraph OUT["Outcomes"]
O1(["Reservation confirmed"])
O2(["Takeout order placed"])
O3(["Manager handoff"])
end
CALLER --> SIP --> STT --> NLU
NLU -->|Lookup| TOOLS
TOOLS <--> CRM
TOOLS <--> CAL
TOOLS <--> KB
NLU --> TTS --> SIP --> CALLER
NLU -->|Resolved| O1
NLU -->|Schedule| O2
NLU -->|Escalate| O3
style CALLER fill:#f1f5f9,stroke:#64748b,color:#0f172a
style NLU fill:#4f46e5,stroke:#4338ca,color:#fff
style O1 fill:#059669,stroke:#047857,color:#fff
style O2 fill:#0ea5e9,stroke:#0369a1,color:#fff
style O3 fill:#f59e0b,stroke:#d97706,color:#1f2937
Discovery dependency: OpenTable sends a meaningful percentage of new guests through its marketplace. Leaving the platform means losing this discovery channel. But the reality is nuanced — studies show that 72% of OpenTable bookings are from guests who already know the restaurant and simply use OpenTable as a booking tool, not a discovery tool.
Phone call anxiety: Operators know they miss calls and fear losing even more reservations if they stop accepting online bookings through platforms. The answer is not "stop offering online booking" — it is "build your own booking channel that actually works."
Guest expectation: Diners have been trained to look for the "Reserve on OpenTable" button. But this is a trained behavior, not a permanent preference. When a restaurant's own website offers easy booking (voice, chat, or web form), guests use it.
Data migration fear: Years of guest data — visit history, preferences, special occasions — lives in OpenTable. Exporting it is possible but operationally daunting.
The system handles inbound phone calls, manages the waitlist, confirms existing reservations, and processes modifications — all without human staff involvement during service hours.
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Inbound Call │────▶│ CallSphere AI │────▶│ Restaurant │
│ (Guest Phone) │ │ Reservation │ │ POS / Book │
│ │◀────│ Agent │◀────│ (Toast, Resy │
└─────────────────┘ └──────────────────┘ │ API, Custom) │
│ └─────────────────┘
│
┌───────────┼───────────┐
▼ ▼ ▼
┌──────────┐ ┌─────────┐ ┌──────────┐
│ Floor Map │ │ Guest │ │ SMS │
│ & Table │ │ Profile │ │ Confirm │
│ Mgmt │ │ DB │ │ System │
└──────────┘ └─────────┘ └──────────┘
from callsphere import VoiceAgent, RestaurantConnector
from callsphere.restaurant import TableManager, GuestDB, WaitlistEngine
# Connect to your reservation system (or use CallSphere's built-in)
restaurant = RestaurantConnector(
pos_system="toast", # or "square", "clover", "custom"
api_key="toast_key_xxxx",
location_id="your_location"
)
# Initialize table management
tables = TableManager(
connector=restaurant,
floor_plan={
"main_dining": {"2tops": 8, "4tops": 6, "6tops": 3, "bar": 12},
"patio": {"2tops": 5, "4tops": 4},
"private_room": {"capacity": 24, "minimum": 10}
},
turn_times={
"lunch": {"2top": 60, "4top": 75, "6top": 90},
"dinner": {"2top": 90, "4top": 105, "6top": 120}
},
buffer_minutes=15 # turnover time between seatings
)
# Guest profile database (owned by the restaurant)
guests = GuestDB(connector=restaurant)
# Configure the reservation agent
reservation_agent = VoiceAgent(
name="Restaurant Reservation Agent",
voice="sophia", # warm, professional
language="en-US",
system_prompt="""You are the reservation host for {restaurant_name},
a {cuisine_type} restaurant in {location}.
Restaurant details:
- Dinner: {dinner_hours}, Lunch: {lunch_hours}
- Capacity: {total_seats} seats
- Private dining available for parties of 10+
- Current wait time for walk-ins: {current_wait}
Your capabilities:
1. Make new reservations (check availability, confirm, send SMS)
2. Modify existing reservations (time, party size, date)
3. Cancel reservations (apply cancellation policy if applicable)
4. Manage the waitlist for same-day seating
5. Answer questions about the menu, dress code, parking, allergies
6. Handle special requests (birthdays, anniversaries, dietary needs)
7. Route large-party and event inquiries to the events team
Conversation standards:
- Greet as: "Thank you for calling {restaurant_name}, this is [name],
how may I help you?"
- Always confirm: date, time, party size, name, phone number
- For parties of 6+, mention that a credit card hold may apply
- For special occasions, ask if they'd like any arrangements
- If fully booked, offer the waitlist or suggest alternative dates
- Never discuss other restaurants or suggest competitors
- Keep the call under 2 minutes for standard reservations
Menu highlights for common questions:
{menu_highlights}""",
tools=[
"check_availability",
"make_reservation",
"modify_reservation",
"cancel_reservation",
"add_to_waitlist",
"check_waitlist_position",
"lookup_guest_profile",
"add_special_request",
"send_confirmation_sms",
"transfer_to_events_manager",
"check_allergen_menu"
]
)
# Handle returning guest recognition
@reservation_agent.on_inbound_call
async def greet_guest(call):
guest = await guests.lookup(phone=call.caller_id)
if guest:
call.set_context({
"guest_name": guest.name,
"visit_count": guest.total_visits,
"last_visit": guest.last_visit_date,
"preferences": guest.preferences, # e.g., "prefers booth, allergic to shellfish"
"upcoming_reservation": guest.next_reservation,
"vip_status": guest.is_vip
})
# Agent opens with: "Welcome back, [name]! It's always
# lovely to hear from you."
waitlist = WaitlistEngine(
table_manager=tables,
notification_channel="sms",
average_wait_accuracy_target=0.85 # within 15% of quoted time
)
@reservation_agent.on_tool_call("add_to_waitlist")
async def handle_waitlist(params):
position = await waitlist.add(
guest_name=params["name"],
party_size=params["party_size"],
phone=params["phone"],
seating_preference=params.get("preference", "any")
)
estimated_wait = await waitlist.estimate_wait(
party_size=params["party_size"],
current_occupancy=await tables.get_occupancy()
)
# Guest receives SMS: "You're #3 on the waitlist at [restaurant].
# Estimated wait: 25-35 minutes. We'll text when your table is ready."
await send_sms(
to=params["phone"],
message=f"You're #{position} on the waitlist at {restaurant.name}. "
f"Estimated wait: {estimated_wait} minutes. "
f"Reply CANCEL to remove yourself."
)
return {
"position": position,
"estimated_wait": estimated_wait,
"confirmation_sent": True
}
For a 120-seat restaurant doing 2 turns per night, 6 nights per week:
| Metric | With OpenTable | With CallSphere AI | Change |
|---|---|---|---|
| Annual reservation platform fees | $75,000-$150,000 | $0 | -100% |
| Annual CallSphere cost | — | $7,200 | — |
| Phone calls answered | 38% | 100% | +163% |
| Reservations from phone/direct | 25% | 72% | +188% |
| Guest data ownership | Platform owns | Restaurant owns | — |
| No-show rate | 12% | 7.5% | -38% |
| Revenue from reduced no-shows | — | $42,000/year | — |
| Average party size (phone booking) | 2.8 | 3.1 | +11% |
| Net annual savings | — | $110,000-$185,000 | — |
The no-show reduction comes from the AI agent's confirmation call sequence: a call 24 hours before and an SMS 2 hours before, with easy rescheduling if plans change. OpenTable's text-only reminders are less effective than a voice confirmation.
Phase 1 — Parallel Operation (Weeks 1-2): Keep OpenTable active. Deploy CallSphere to handle phone calls that previously went to voicemail. This immediately captures lost reservations without disrupting the existing channel. Track how many phone-originated bookings the AI captures.
Still reading? Stop comparing — try CallSphere live.
See the restaurant AI agent handle a real call — complete, industry-specific, and live in your browser. No signup.
Phase 2 — Direct Channel Promotion (Weeks 3-6): Add "Call to Reserve" prominently to your website, Google Business profile, and social media. Update your outgoing voicemail to reference the AI booking line. Begin tracking what percentage of your OpenTable bookings are from repeat guests who already know your restaurant (these guests can be migrated to direct booking).
Phase 3 — OpenTable Tier Reduction (Month 2-3): Downgrade your OpenTable subscription to the basic tier. Remove premium placement. Monitor whether reservation volume decreases — if most of your OpenTable traffic was repeat guests who now book direct, the impact will be minimal.
Phase 4 — Full Independence (Month 4+): For restaurants where the data confirms that OpenTable was primarily a booking tool (not a discovery channel), cancel the platform entirely. Redirect the saved fees into direct marketing, Google Ads, and guest experience improvements that drive word-of-mouth.
A farm-to-table restaurant in Portland with 80 seats deployed CallSphere's reservation agent as a complete OpenTable replacement. After 6 months:
This is the most common concern, and it is often overstated. Analyze your OpenTable data: what percentage of bookings come from guests who searched for your restaurant by name versus those who discovered you through OpenTable's marketplace? For most established restaurants, 65-80% of OpenTable bookings are name searches — these guests already know you. The remaining 20-35% who discover you through OpenTable can be replaced through Google Business optimization, Instagram, and targeted local ads at a fraction of the cost.
Yes. CallSphere's guest profile database stores seating history. When a returning guest calls, the agent can reference their previous table assignment: "Last time you sat at the corner booth in the main dining room. Would you like to request that table again?" This level of personalization actually exceeds what most human hosts can recall for non-VIP guests.
The agent detects the caller's time zone from their area code and confirms reservation times in the correct zone. If someone from the East Coast calls a West Coast restaurant and asks for "dinner at 7," the agent clarifies: "That would be 7 PM Pacific Time — is that correct?" Language switching is automatic — CallSphere supports 30+ languages with native-quality voice synthesis, which is particularly valuable for restaurants in tourist-heavy areas.
The agent handles high-volume periods without degradation. On Valentine's Day or New Year's Eve, when a restaurant might receive 200+ calls, the AI agent handles them all simultaneously. It can manage priority access for VIP guests, enforce special event pricing and menu requirements, collect deposits for premium seatings, and maintain a waitlist when all time slots are booked. The system also sends automated "availability alert" messages to waitlisted guests when cancellations open spots.
CallSphere recommends a 60-90 day parallel operation where both systems run simultaneously. Phone calls route to the AI agent while OpenTable continues handling online bookings. This gives the restaurant data on how many reservations the AI captures, what the guest experience is like, and whether any issues need tuning before reducing reliance on the third-party platform. No reservations are lost during the transition because both channels remain active.
Written by
Sagar Shankaran· Founder, CallSphere
Sagar Shankaran is the founder of CallSphere, where he builds production AI voice and chat agents deployed across healthcare, hospitality, real estate, and home services. He writes about agentic AI, LLM engineering, and shipping voice agents that handle real calls in production.
See how AI voice agents work for your industry. Live demo available -- no signup required.
A how-to for Colombian education and tutoring SMBs to answer parents and students instantly, book trial classes 24/7 in Spanish and English, and grow enrollment with a CallSphere AI agent.
Tbilisi professional-services firms serving relocating founders and IT companies use CallSphere AI voice and chat agents to answer enquiries 24/7 in English, Georgian and Russian and book consultations.
A practical how-to for Palau eco-resorts and dive operators on capturing every high-value, multilingual enquiry with a CallSphere AI voice and chat agent, while honouring Palau’s marine-conservation commitments.
How salons, spas and wellness SMBs across the UAE, Saudi Arabia and Qatar use CallSphere AI voice and chat agents to capture every booking 24/7 in Arabic, English and expat languages, and cut no-shows.
How estate agents and property managers in Luxembourg City and across the Grand Duchy use CallSphere to capture multilingual viewing and enquiry calls 24/7, GDPR compliant.
A practical guide for Senegalese logistics, freight, legal and consulting SMBs in Dakar and Thiès to capture every enquiry 24/7 with a CallSphere AI voice and chat agent in French, Wolof and English.
© 2026 CallSphere LLC. All rights reserved.
Made within New York
Watch how CallSphere handles real customer calls, schedules appointments, and processes payments — live.
Try Live DemoBook a DemoCalculate Your ROI