Numa vs Toma vs 11Sight: Which AI Voice Agent Is Best for Your Car Dealership in 2026?

Numa vs Toma vs 11Sight: Which AI Voice Agent Is Best for Your Car Dealership in 2026?

AI voice agent market for automotive dealerships has matured rapidly. Two years ago, the conversation was about whether AI could answer a phone call. Today, the conversation is about whether AI can run your dealership's entire customer-facing operation - and whether you can measure, monitor, and improve it over time.

That shift matters. As Shub Agarwal, Professor of AI at USC and Founder of AI Trust Lab, frames it: AI product creation is going through the same shift aviation did - from early experiments held together by hope and duct tape to product discipline and frameworks that actually fly. The dealerships winning with AI in 2026 aren't the ones who picked the flashiest demo. They're the ones who picked the platform with the strongest operational framework underneath.

This comparison evaluates the three leading AI voice platforms for automotive - Numa, Toma, and 11Sight - across four pillars that matter to dealership operations: Customer Experience, Dealer Experience, Management Analytics, and Solution Architecture.

1: Customer Experience

This is what your customers feel. Voice quality, speed, accuracy, personalization - the factors that determine whether a caller stays on the line or hangs up and calls your competitor.

Voice Quality

• Numa: Reliable voice quality through their proprietary stack. Natural enough for routine interactions, though their primary innovation focuses on unifying communication channels rather than pushing the boundaries of voice naturalness.
• Toma: Strong emphasis on voice personalization. Their "One-Size-Fits-One" approach trains a custom voice per dealership, mimicking the tone of your best employee. Callers get a voice that feels familiar to your brand.
• 11Sight: High quality with multiple personalized voice options with emotional intelligence - the AI adjusts tone, pacing, and warmth based on the caller's emotional state, not just the words being spoken. 

Latency

• Numa: Standard response times. Adequate for most interactions but not a stated differentiator.
• Toma: Markets sub-300ms response times - among the fastest in the market. Speed is a core part of their value proposition.
• 11Sight: Sub-second response times. ADK delivers consistently low latency at enterprise scale without degradation across high call volumes.

All three platforms deliver acceptable latency for natural conversation. Toma leads on raw speed claims. 11Sight's advantage is maintaining low latency at scale through Google's infrastructure.

Call Resolution Rate

• Numa: Does not publicly report autonomous call resolution rates. Their model emphasizes routing and collaboration rather than autonomous resolution.
• Toma: Does not publicly report autonomous resolution rates. Their focus is on deeply customized handling with smart escalation when needed.
• 11Sight: Reports 60%+ first-call resolution rate - meaning six out of ten service calls are handled entirely by AI without any human involvement. This figure was confirmed on the Google Cloud blog, showing improvement from 40% in February to 60% by November.

This is a significant operational metric. A 60% resolution rate means your human staff focuses only on the 40% of calls that genuinely need them.

Ability to Provide Information in Sequence

This is an underappreciated capability. Can the AI walk a customer through a multi-step process - explaining a recall, confirming their VIN, describing the repair, offering appointment slots, and booking - in a logical sequence without losing context?

• Numa: Handles structured conversations through its Smart Inbox workflow, but the AI's primary role is routing and flagging rather than conducting multi-step autonomous conversations.
• Toma: Performs well on sequential tasks like service scheduling. Their VIN-specific logic can walk through OEM-recommended services by mileage in a natural sequence. Strong on structured service conversations.
• 11Sight: Each specialized agent is designed for end-to-end sequential task completion. The Service Agent, for example, greets by name, retrieves vehicle history, checks bay availability, suggests the right service based on mileage and history, offers time slots, books the appointment, and sends confirmation - all in one unbroken conversation flow.

Customer History and Contextual Memory

• Numa: Connects to DMS systems and aggregates history into the Smart Inbox. Context is shared across team members - strong for internal collaboration, though the AI's use of this data in live conversations is more limited.
• Toma: VIN-specific logic pulls vehicle details and recommends service packages by mileage. The agent is trained on each dealership's specific data and workflows, providing relevant context during calls.
• 11Sight: Total access to every past interaction - last call, last appointment, scheduling preferences, open recalls, and previous conversation context. The AI uses this data to personalize every conversation from the first second. A returning customer is greeted by name with reference to their specific history before they say a word.

Overall Customer Satisfaction Impact

• Numa: Reports CSI score improvements of 29 points at specific customer dealerships - an impressive figure tied to their communication visibility improvements.
• Toma: Customer testimonials consistently describe the experience as "like having another team member." Satisfaction improvements are qualitative rather than platform-wide metrics.
• 11Sight: We are keeping track of CSI score improvements
•  with 28% more appointments booked and 10% of those booked during off-hours when no human would have been available.

2: Dealer Experience (Operational)

This is what your team experiences day to day. How does the AI integrate into your staff's workflow? What happens when a call needs to be transferred? How does the system communicate with your team?

Call Context and Meeting Goals as Part of Transfer

When the AI transfers a call to a human, does your team get context - or do they start from scratch?

• Numa: The Smart Inbox provides shared context across team members. When a call is routed, the receiving team member can see the conversation history, prior messages, and relevant notes. Strong for team-wide visibility.
• Toma: Their "Transfer Clawback" feature is thoughtful - if a transferred call isn't answered by a human, the AI pulls the call back, apologizes, and offers alternatives. When transfers do connect, the team receives context and follow-up alerts via Inbox.
• 11Sight: Every transfer includes the full conversation context - who the customer is, what they drive, why they're calling, what has already been discussed, and what the customer's emotional state is. The receiving team member sees all of this before they pick up.

Email / SMS Notifications

• Numa: Strong multi-channel communication. The Smart Inbox unifies SMS, voice, and internal messaging. Team notifications are a core feature.
• Toma: Automates voice and SMS interactions. Missed call recovery via text is a standout - when a customer hangs up or disconnects, Toma follows up with a text immediately.
• 11Sight: Full email and SMS capabilities across all agents. The Outbound Agent handles interactive SMS and voice campaigns for recalls, reminders, sales, and confirmations. Internal team notifications for urgent outcomes.

Integrated Call Center

• Numa: Functions as a communication layer on top of existing operations. The Smart Inbox acts as a unified workspace but does not replace or provide a built-in call center infrastructure.
• Toma: The Inbox product serves as a centralized workspace for managing AI-handled calls and human follow-ups. Functions more as a visibility layer than a full call center.
• 11Sight: Includes a built-in call center where dispatch and transfer to human team members is managed directly within the platform. The AI agents and human agents operate within the same system - not as separate tools stitched together.

3: Dealer Experience - Management

This is what your GM, service director, and ownership group need. Not just "is the AI working?" but "how well is it working, where is it improving, and can I prove ROI?"

Detailed Dashboard

• Numa: Offers real-time manager visibility through the Smart Inbox and LiveCSI product. Managers can monitor responsiveness, sentiment, and team performance. Dashboard-level insights are a core selling point.
• Toma: Provides call intelligence with transcription, analysis, and key detail reporting. Their dashboard shows what the AI handled versus what needs human attention. Straightforward and functional.
• 11Sight: The Sentinel Agent generates two types of dashboards: external dashboards summarizing operational performance for management (what was accomplished, how many calls resolved, appointment conversion rates), and internal dashboards used by engineering and QA to improve AI performance. This dual-dashboard approach separates the "what happened" view for executives from the "how to improve" view for operations.

Consistent Vocabulary: Handled, Dispatched, and Resolved

This matters more than it sounds. If your AI platform doesn't use consistent terminology for call outcomes, your reporting is unreliable. What counts as "handled"? Is a transfer "resolved" or "dispatched"?

• Numa: Uses its own terminology within the Smart Inbox ecosystem. The vocabulary is consistent within Numa but may not align with standard dealership KPI definitions without configuration.
• Toma: Tracks outcomes clearly - appointment booked, callback requested, transferred, etc. Clean and functional for core use cases.
• 11Sight: Uses a consistent vocabulary across all agents: calls are classified as resolved (AI completed the task autonomously), dispatched (transferred to a human as planned), escalated (transferred because the AI couldn't handle it), or abandoned (customer hung up). Every conversation is instrumented with structured data - what information was provided, what was gathered, what tasks were accomplished, whether a human was needed, and whether the customer hung up prematurely. This structured approach makes cross-dealership benchmarking possible.

CSI Analytics

• Numa: LiveCSI is purpose-built for this - it surfaces customer satisfaction risks in real time and flags hot cases for manager intervention. Among the strongest CSI monitoring tools in the market.
• Toma: Provides follow-up alerts for frustrated customers and quality signals, but CSI analytics is not a standalone product feature.
• 11Sight: The Sentinel Agent continuously monitors every conversation for customer satisfaction signals - frustration, confusion, delight - and feeds this data into both real-time alerts and historical CSI analytics. The system tracks satisfaction trends over time, not just individual call flags.

Customized Metrics Aligned to Dealer KPIs

• Numa: The Smart Inbox and Opportunities products provide metrics around response time, conversion, and demand capture. Customization to specific dealer KPIs requires configuration.
• Toma: Metrics focus on core operational outcomes - calls handled, appointments booked, time saved. Straightforward and dealership-relevant.
• 11Sight: The platform records every conversation with detailed instrumentation and supports customized reporting aligned to each dealership's specific KPIs - whether that's first-call resolution, appointment conversion rate, recall completion percentage, or after-hours capture rate. The management dashboard can be configured to show the metrics each dealer group prioritizes.

Month-over-Month Improvement Analysis

• Numa: Provides performance tracking over time through its dashboard. Trends are visible but are primarily focused on communication metrics.
• Toma: Tracks performance improvements but doesn't publicly highlight MoM improvement as a structured capability.
• 11Sight: The Sentinel Agent's continuous learning loop creates a built-in improvement cycle. Every call generates data that feeds back into agent training. The management dashboard shows MoM trends across all key metrics - resolution rates, satisfaction scores, escalation rates, and appointment conversion. This creates accountability and a clear improvement trajectory that dealership leadership can track and report to ownership.

Ability to Dynamically Change Agent Priorities

Can you tell the AI "this week, push tire promotions" or "prioritize recall outreach over service reminders"?

• Numa: The Opportunities product identifies hidden demand, but dynamic priority-shifting of AI behavior mid-campaign is not a highlighted capability.
• Toma: Custom agent training means priorities can be adjusted, but changes typically require reconfiguration rather than a real-time priority toggle.
• 11Sight: The platform includes interfaces that allow daily adjustment of agent task priorities. If you're running a tire promotion this week or need to surge recall outreach before a deadline, you can shift what the AI prioritizes in conversations - which offers it leads with, which services it recommends, which campaigns get outbound attention - without retraining the model.

4: Solution Architecture

This is where the long-term bet gets made. The underlying architecture determines not just what the platform can do today, but what it can become.

Voice AI Agent Delivery

• Numa: Delivers a unified communication AI layer. The Operator product serves as an AI receptionist. The architecture is mature and well-integrated with major DMS platforms (CDK, Reynolds, Tekion, Xtime). 1,200+ dealerships validate the production readiness. Backed by $55M in funding including Google's Gradient Ventures.
• Toma: Delivers a single, deeply customized voice AI agent per dealership. Proprietary voice AI infrastructure with SOC 2 Type II compliance. YC graduate, backed by a16z with $17M raised. Has processed 1M+ dealership calls. The "One-Size-Fits-One" philosophy means each deployment is tailored - but it's one agent, not a multi-agent system.
• 11Sight: Delivers five specialized AI agents running simultaneously on Google's Agent Development Kit (ADK) on Vertex AI. Each agent handles a distinct function (service, front desk, sales, outbound, and quality monitoring). Agents can pass calls to each other collaboratively. The platform also creates simulation AI agents that call in and act like customers - used for testing, training, and QA before and after deployment. This multi-agent architecture is fundamentally different from a single-agent or communication-layer approach.

Sentinel AI Agent

This is the architectural differentiator that has no equivalent on any competing platform.

• Numa: No dedicated AI supervisory agent. LiveCSI provides dashboard-level satisfaction monitoring - valuable, but it operates at the reporting layer, not the conversation layer.
• Toma: No dedicated AI supervisory agent. Transfer Triggers and Safeguards provide rule-based escalation - effective for defined scenarios, but not continuous AI-driven quality monitoring across every conversation.
• 11Sight: The Sentinel Agent is a dedicated AI supervisory agent trained on thousands of real dealership interactions. It monitors every live conversation across all other agents in real time. It detects quality issues, emotional signals, compliance risks, and escalation triggers - and acts on them instantly. Post-call, it generates structured data for both external management dashboards and internal engineering QA dashboards. The Sentinel doesn't just watch - it drives continuous improvement by feeding learning data back into every other agent.

This is the difference between a platform that answers calls and a platform that answers calls, monitors its own quality, and gets measurably better every month.

So, Which One Is Right for Your Dealership?

The AI voice agent market has moved past the "does it work?" phase. The question now is which platform matches how you actually want to run your dealership.

Choose Numa if your primary challenge is communication fragmentation. You need one place to see every customer conversation - voice, text, messaging - and you want a mature platform with the largest installed base and strong DMS integrations. Numa's Smart Inbox and LiveCSI are best-in-class for communication visibility and team collaboration.

Choose Toma if you want a single, deeply personalized AI agent deployed fast. Toma's "One-Size-Fits-One" approach delivers a voice that sounds like your dealership from day one, with strong service scheduling and smart escalation. If your core need is handling service calls and missed call recovery with a tailored agent, Toma delivers.

Choose 11Sight if you need a complete operational framework - not just an AI that answers calls, but a system that runs five specialized agents across every department, monitors its own quality in real time through the Sentinel Agent, provides dual dashboards for management and QA, lets you dynamically adjust priorities, and gets measurably better every month. Built on Google's ADK and Vertex AI, 11Sight is the platform for dealerships that want product discipline and frameworks - not just demos, experiments, and hope.

The right choice depends on where you are. If you're solving one problem, any of these platforms can help. If you're building the operating system for how your dealership handles every customer interaction - and you want to prove ROI to your ownership group with real data - the architecture underneath matters as much as the voice on the phone.

Ready to see the difference? Visit www.11sight.com to book a demo.