Service to display compact summaries of differences between two vehicles?

[Viracocha]

I want to build a page that allows a user to select a vehicle and compare the "top" (I understand that this may be a completely subjective differentiator) enhancements or differences between two vehicles? For example, if a shopper is driving a 2018 Crew Cab XLT F150 and is looking to see what the main differences are between their current vehicle and the 2025 CrewCab XLT F150 are, they could see the top 5 changes between the two model years broken out in standard groups (Convenience, Safety, Powertrain etc)
I played a bit with this in ChatGPT but found it tough to get it to provide summaries that would be compelling for a shopper to see value in reading. I know that I could create these tables manually in a few weeks of work but was hoping someone else has already done it...
Do any of the larger Automotive Data providers have their data structured in any way that could be utilized to achieve this? Or has anyone seen this done successfully in the past using AI?

**Edit - Perhaps I need to talk to Uncle Joe, just reading his Automagic thread and it seems like he's gone down this road?!

 

[Chris Cachor]

Your best bet is going down the AI path

 

[Carsten]

This?

To do that on a website, you're going to need an api call to chatGPT and some other things.
If would ask chatgpt next to create a program to provide this data.

 

[joe.pistell]

you're going to need an api call to chatGPT

NOTE: AI systems are NOT REAL TIME, they lag about 1yr or more behind today's info

Ask your AI for it's "knowledge cutoff" date.​

so, the 2025 data is an "hallucination" (aka A wild ass guess).

Other AI svcs:

• Claude 3.0: April 2024.
• Gemeni Advanced: Real Time (no cut off)
• Perplexity: Real Time (no cut off)
• Meta: December 2023
• X: Never use it, I don't care

 

[joe.pistell]

• if a shopper is driving a 2018 Crew Cab XLT F150 and is looking to see what the main differences are between their current vehicle and the 2025 CrewCab XLT F150

• Do any of the larger Automotive Data providers have their data structured in any way that could be utilized to achieve this? Or has anyone seen this done successfully in the past using AI?

**Edit - Perhaps I need to talk to Uncle Joe, just reading his Automagic thread and it seems like he's gone down this road?!

Ahh, Viracocha is displaying excellent JTBD software design. JBTD is Job To Be Done. Google it, it's powerful and so simple.

Yes, auto data providers have highly structured, VIN specific data. Infact, it comes in 'levels' of perfection (or completeness). The more you spend, the more complete is the record.

In your example (2018 vs 2025 F150 XLT), AI can do a OEM YMMT based 'generic' comparison (you'll need AI to be real time) This is helpful in high level research and is seen in SEO landing pages.

Using the JTBD model, the shopper is on the 2025 CrewCab XLT F150 VDP likes what they see and NOW they want a simple way to compare the car they own vs the car they want to buy. This task/solution 'should be' the websites responsibility, but, the JTBD philosophy is alien to these ppl, so, they can't see it.

That being said, this is a perfect JTBD by an intelligent chat bot on the dealer's site. THe bot would need to be aware of the page the shopper is on (Carvana's bot does this) and if the bot has this info, the bot could ask the shopper the YMMT the shopper has. The bot could ask the shopper if they'd give the bot the VIN or Plate#, they'd have the build sheet data for that specific car and it would improve the quality of the comparison (p.s. tradein forms harvest this data and chat bots should have this info if provided).

Viracocha, nice rabbit hole you've found! Very helpful for shopper knowledge. Too bad 99.9% of the software ppl in autoare chasing lead gen.

 

[Chris Cachor]

Yea to what Joe is saying - as part of your API call, you would be passing in data as context for the prompt to compare against. As part of your prompt, you also want to include details of what specific data you'd like compared and how detailed or terse the response to be. You likely could use existing AI playgrounds and then paste in output from VIN decoding services to see how responses vary. I would stray away from having the output be a text dump of "specs" but rather focus on things like "truck of the year", "top safety picks", etc. People buying cars don't really care about gross vehicle weight or torque numbers. They do care about reliability, safety, and popular features like remote start, CarPlay, adaptive cruise-control, engine start-stop. Some people find the engine start stop features annoying and would rather a vehicle not having the feature (like me, lol).

 

[Carsten]

Or you could skip the AI and do api calls to services that provide data on the vehicles.