Beyond the Glovebox: Turning the "Check Engine" Light into an Instant Answer
Multimodal Knowledge Agent
It is a universal moment of panic: You are driving, and a mysterious red symbol flashes on your dashboard.
Is it critical? Can you drive home? Or do you need to pull over immediately?
In the past, the driver had two bad options:
Dig through the Glovebox: Try to find the physical owner's manual, then flip through the index to find "Icon with exclamation point inside a gear."
Call the Dealership: Waste the service center's time (and the driver's time) with a frantic phone call for a simple question.
Car manuals are dense, technical, and often outdated the moment they are printed. As vehicles become more software-defined, the gap between the car's complexity and the driver's understanding grows.
At Evonence, we are bridging this gap with the Interactive Owner's Manual Agent.
The Solution: Don't Type It, Just Snap It
We replace the static PDF manual with a Multimodal Knowledge Agent that lives in your brand's mobile app.
Instead of typing "weird red circle light" into a search bar, the driver simply opens the app and snaps a photo of the dashboard.
Driver: [Uploads photo of dashboard] "What is this light and can I keep driving?"
Agent: "That is the Tire Pressure Monitoring System (TPMS) warning. It indicates your rear left tire is significantly under-inflated. Since the light is solid (not flashing), you can drive cautiously to the nearest air pump, but you should not drive at highway speeds."
The agent identifies the specific icon, retrieves the exact page from the digital manual, and summarizes the safety advice in plain English.
Under the Hood: The Power of "Vision" AI
This isn't keyword search. This is Computer Vision powered by Generative AI.
1. Gemini 3 Flash (Multimodal Input)
We leverage Gemini 3 Flash because it is natively multimodal. It doesn't just process text; it processes pixels. It can distinguish between a "Check Engine" light and a "Low Oil" light even in low-light conditions or at an odd angle.
2. Vertex AI Search (Grounded Truth)
Generative AI can hallucinate, which is dangerous in automotive safety. That is why we "ground" the model using Vertex AI Search. The agent is strictly constrained to answer only based on the official technical documentation you provide. If the manual doesn't say it, the agent doesn't guess.
The Migration Advantage: Why Text-Bots Fail
Competitors often try to solve this with text-based chatbots (like AWS Lex or early OpenAI wrappers). These fail because drivers don't know the technical name of the part they are looking at.
You cannot search for "Catalytic Converter Failure" if you don't know that is what the icon means.
By migrating to a Google Cloud Multimodal approach, you remove the language barrier. The user provides the visual reality; the AI provides the technical definition. This drastically increases adoption and successful self-service rates.
The Evonence Approach: A 4-Week "Service Deflection" Pilot
This is one of the highest-ROI "Quick Wins" in the automotive sector.
Week 1: Ingest your PDF manuals and technical bulletins into Vertex AI.
Week 2: Configure Gemini's vision prompts for dashboard recognition.
Week 3: Build the mobile-friendly chat interface.
Week 4: Launch to a test group of owners.
The Result: Deflect up to 30% of simple inbound service calls, freeing up your service advisors to handle actual repairs, not FAQs.
Give your drivers the ultimate co-pilot. Contact Evonence to transform your owner's manual today.
