Project Overview
| Client Industry | Automotive (Multi-Location Dealership Group) |
| Business Type | Regional auto dealership group operating 6 locations across 3 brands |
| Project Duration | 16 weeks |
| AI Service Provided | Web Development |
| Technologies Used | .Next.js, TypeScript, Node.js, OpenAI GPT-4 class models, PostgreSQL, AWS (ECS, S3, CloudFront), DealerSocket CRM API, vAuto inventory API, Twilio |
The Client Challenge
The dealership group operated 6 locations across 3 vehicle brands, but each location’s website had been built separately over the years by different third-party dealer website vendors, resulting in inconsistent experiences, poor lead quality, and no unified digital presence across the group.
The operational and marketing challenges:
- Each of the 6 dealership websites ran on a different vendor platform with inconsistent design, functionality, and inventory data freshness inventory shown online frequently didn’t match what was actually on the lot.
- Website-generated leads were low quality; a large share of online inquiries were vague “more info” form submissions that sales staff found difficult to act on quickly, and conversion from web lead to actual showroom visit was well below the group’s internal benchmark.
- None of the six sites gave customers meaningful help narrowing down which vehicle actually fit their needs and budget just a standard inventory search and filter interface similar to every other dealer website.
- Leads submitted through the websites took, on average, over an hour to reach a sales rep due to the disconnected lead routing between each site’s vendor platform and the dealership group’s DealerSocket CRM.
- Group leadership wanted a consistent, modern brand presence across all 6 locations while preserving each location’s specific inventory and local identity, something the current fragmented vendor approach made difficult to achieve.
- The group’s marketing budget for paid search and social advertising was being undermined by poor on-site conversion, meaning qualified ad traffic was arriving at sites that weren’t converting it effectively.
Our Solution
Air Brite Labs built a unified web platform for all 6 dealership locations, combining consistent modern design and real-time inventory accuracy with an AI-powered vehicle matching assistant designed to improve both lead quality and the customer’s actual car-shopping experience.
Key capabilities delivered:
- Unified, consistent platform across all 6 locations — each dealership retains its own branding and local inventory, but all sites now run on the same modern, fast, consistently functioning platform rather than 6 disconnected vendor systems.
- Real-time inventory accuracy — vehicle inventory syncs directly from the group’s vAuto system, eliminating the stale listing problem that had been frustrating customers who arrived to find advertised vehicles already sold.
- AI-powered vehicle matching assistant — customers describe what they need (“a reliable family SUV under $35k with good safety ratings”) and get matched to actual available inventory across the group’s locations, rather than navigating generic filter menus.
- Instant, routed lead capture — leads generated through the matching assistant include rich context (vehicle interest, budget, specific questions) and route immediately to the correct location’s sales team through direct DealerSocket CRM integration, cutting the lead response delay.
- SMS-based immediate follow-up — high-intent leads trigger an automatic SMS to the customer confirming receipt and setting expectations for sales follow-up, reducing the “did my inquiry even go through” uncertainty that was hurting conversion.
- Cross-location inventory visibility — customers interested in a vehicle at one location can see relevant inventory at the group’s other locations, capturing cross-location sales opportunities the fragmented site structure had been missing entirely.
Technical Approach
Frontend Architecture: Built with Next.js and TypeScript as a single platform serving all 6 dealership sites through location-specific routing and theming, giving the group consistent performance and functionality while preserving each location’s distinct brand presentation.
Inventory Integration: Real-time synchronization with the group’s vAuto inventory management system via API, replacing the batch-updated or manually maintained inventory data that had been causing the stale listing problem across the prior vendor sites.
AI Vehicle Matching: The matching assistant uses OpenAI GPT-4 class models to interpret natural-language customer descriptions of their needs and match them against real-time inventory data across all 6 locations, grounded strictly in actual available vehicles rather than generating generic recommendations.
CRM Integration: Built a direct integration with Dealer Socket CRM via its API, so leads generated through the website including the rich context captured by the AI matching conversation flow immediately into the correct location’s sales workflow rather than requiring manual re-entry or routing.
Lead Follow-Up: Twilio integrated for automated SMS confirmation to customers immediately after lead submission, addressing the response-delay uncertainty that had been contributing to poor lead-to-visit conversion.
Backend: Node.js API layer orchestrates the inventory sync, matching assistant logic, and CRM lead routing, with PostgreSQL storing structured vehicle and lead data shared consistently across all 6 location sites.
Infrastructure: Deployed on AWS ECS with CloudFront for fast content delivery and S3 for vehicle imagery and media assets, built to handle the combined traffic of all 6 previously separate dealership sites on unified, consistently performant infrastructure.
Implementation Process
- Discovery & Requirement Analysis (Weeks 1–2): Audited all 6 existing dealership sites, interviewed sales staff at multiple locations about lead quality issues, and mapped the group’s vAuto and DealerSocket data structures for integration planning.
- Prototype / PoC (Weeks 3–4): Built a working prototype of the AI vehicle matching assistant for one location, validated against real inventory data and tested with sales staff for lead quality and relevance.
- Development (Weeks 5–10): Built out the full unified platform architecture, location-specific theming for all 6 sites, and the complete matching assistant and lead routing system.
- Integration (Weeks 11–12): Connected the vAuto inventory sync and DealerSocket CRM integration across all 6 locations, integrated Twilio for automated lead confirmation.
- Testing (Week 13): Ran a soft launch for 2 of the 6 locations, comparing lead quality, response time, and conversion against the prior vendor sites before expanding to the remaining locations.
- Deployment (Weeks 14–15): Phased rollout to the remaining 4 locations, retiring each prior vendor site as its replacement went live.
- Optimization (Week 16 and ongoing): Refined the matching assistant’s conversational guidance based on real customer interactions and tuned CRM lead routing rules based on sales team feedback.
Key Features Delivered
- Unified, modern web platform serving all 6 dealership locations with consistent performance
- Real-time inventory synchronization via vAuto integration
- AI-powered conversational vehicle matching assistant grounded in actual available inventory
- Direct DealerSocket CRM integration for immediate, context-rich lead routing
- Automated SMS lead confirmation reducing customer response-time uncertainty
- Cross-location inventory visibility surfacing sales opportunities across the group
- Consistent local branding preserved across each dealership’s distinct identity
Business Results
- Lead-to-showroom-visit conversion rate improved measurably across the dealership group following the platform launch
- Average lead response time dropped from over an hour to near-immediate, driven by direct CRM integration and automated SMS confirmation
- Stale inventory complaints were effectively eliminated, following the real-time vAuto synchronization replacing the prior batch-updated listings
- Leads generated through the AI matching assistant were rated meaningfully higher quality by sales staff compared to generic contact form submissions, based on internal sales team feedback
- Cross-location vehicle inquiries increased, capturing sales opportunities the prior fragmented site structure had been missing
- Paid advertising return on ad spend improved, attributed to better on-site conversion from the same qualified traffic volume
Technology Stack
| Layer | Technology |
|---|---|
| Frontend | Next.js, TypeScript |
| Backend | Node.js |
| AI / Conversational Matching | OpenAI GPT-4 class models |
| Database | PostgreSQL |
| Inventory Integration | vAuto API |
| CRM Integration | DealerSocket API |
| Communication | Twilio |
| Cloud Infrastructure | AWS ECS, S3, CloudFront |
Why the Solution Worked
The platform succeeded because it addressed the actual source of poor lead quality generic contact forms and stale inventory rather than simply giving the dealership group a visual refresh of the same fundamental website structure every competitor used. Grounding the AI matching assistant strictly in real, current inventory data was critical for a car-buying context, where a customer excited about a vehicle that turns out to be unavailable creates real frustration and damages trust in the dealership.
Unifying all 6 locations onto one platform while deliberately preserving each location’s distinct local branding also mattered it gave the group real engineering and marketing efficiency without asking individual dealership managers to give up the local identity that mattered to their specific market and customer base.
Future Scalability
The platform is designed to support the dealership group’s continued growth, including:
- Extending the AI matching assistant to support trade-in value estimation as part of the vehicle matching conversation
- Adding a service scheduling module integrated with each location’s service department systems
- Onboarding additional dealership locations onto the unified platform as the group continues to grow
- Building more sophisticated cross-location inventory recommendation logic as the group’s combined inventory data accumulates
Because the platform is built as a unified architecture with location-specific configuration rather than 6 separate codebases, adding new locations or capabilities extends the existing system rather than requiring a rebuild per location.
Final Outcome
The dealership group replaced 6 fragmented, inconsistent vendor websites with one unified, fast, AI-enhanced platform that generates meaningfully better leads and gets them to sales staff faster while preserving the local brand identity that matters to each individual dealership’s market.


