In our recent Keyloop Drivetime podcast, our host Jacqui Barker spoke to Andy Gray, CEO and co-founder of Kortical. This episode serves to cut through the noise around AI and get practical about what actually works in automotive retail.
However, (and this is a big one), only 8% of AI projects succeed, and the average time to production is 36 months. That’s a daunting prospect for dealers and OEMs. Thankfully, Andy shared how to break that pattern by starting small, focusing on quick wins, and building on proven approaches instead of attempting big-bang transformations. In fact, Andy comments, “We started building on AI when no one really knew what it was, and suddenly customers were asking for it.”
We’ve pulled together some key takeaways from this episode. It’s packed with automotive retail insights for anyone looking to bridge the gap between cutting-edge tech and real-world application. Here’s what Jacqui and Andy discussed:
- AI vs machine learning: The difference explained simply.
- Why most AI projects fail: And how to shortcut success with proven tools.
- The goldmine of automotive data: From service records to finance. agreements, there’s more value in your data than you think.
- Personalisation is the new expectation: Customers now expect an Amazon-style experience – right car, right time.
- Human + AI partnership: AI as an assistant that frees salespeople to focus on relationships, not admin.
AI vs machine learning
Andy explained the moving goalposts of AI.
As soon as something is well understood, people stop calling it AI and give it a new name, like optical character recognition.
He traced the origins back to John McCarthy in the 1950s, framing AI as “anything brain-inspired”.
Machine learning, by contrast, is a specific method: instead of coding thousands of rules by hand (e.g. for facial recognition), you let algorithms learn patterns from data. That shift is what made modern breakthroughs possible; from OCR to today’s large language models.
Why many AI projects fail, and how to avoid it
An 8% success rate, with an average of 36 months to reach production.
Multiply those industry figures and you get one successful project every 20 years if you approach it the wrong way.
Failures often occur at the translation layer. Businesses don’t know how to turn their messy data into machine-readable inputs and back into business value. (This is exactly why improving data quality in dealerships matters so much. See how Keyloop and AI Assistant are tackling it here).
His advice? Don’t go for big-bang programmes. Instead:
- Start small with one process that’s repetitive and structured (digital in, digital out).
- Use proven solutions already working in similar contexts.
- Incremental wins build confidence and ROI and give you the roadmap for scaling AI effectively.
The goldmine of automotive data
Andy challenged the myth that you must centralise all data before using AI. Instead, he urged retailers to start with silos of data already available. For example:
- Renewals AI can predict when a customer is likely to buy again based on agreement info alone.
- With just a subset of data, Kortical achieved an average 7.8% uplift in personalised outreach.
He emphasised that automotive has a ton of data spanning service, finance, and marketing. The real challenge is surfacing and activating it, not hoarding it in one big project. For a broader view of why clean, connected data underpins AI success in retail, see our take on AI in Automotive and Beyond.
Personalisation is the new expectation
Andy explained that people no longer spend hours scrolling through endless options or visiting multiple review sites. Instead, they increasingly trust brands to serve them the right choice, at the right time.
For automotive, that means retailers can no longer rely on generic marketing or one-size-fits-all offers. AI can analyse signals such as finance agreements, service history, or even lifestyle changes, and recommend the most relevant next step. Whether it’s a renewal, a service booking, or a tailored deal.
When done well, personalisation keeps customers in your ecosystem, builds trust, and reduces the chance they’ll drift off to explore competitors. As Andy put it:
If people believe your brand consistently understands their needs, they’ll stop looking elsewhere.
Human + AI partnership
Andy stressed AI should amplify people, not replace them. Examples from the episode include:
- An AI chatbot trained on a retailer’s brand voice can answer queries in a natural, curated way; a far cry from old “pick option A, B, C” bots.
- Bots connected to the DMS can qualify leads, check real-time stock, and even book test drives before a salesperson is free.
- Salespeople gain their own “assistant” to handle admin and outreach, so there’s more time for human connection and value-building.
Today’s consumer is conditioned by the personalisation we see everywhere now. If automotive can replicate that seamless, personalised journey, loyalty and renewals will follow.
Final takeaways
AI has moved far beyond the buzzwords. As Andy highlighted in this Drivetime episode, success isn’t about chasing the biggest, flashiest projects, it’s about making AI practical. The automotive industry has an abundance of untapped data, and the opportunity lies in starting small, using what’s already available, and building momentum through measurable wins.
Retailers who adopt this approach will be best placed to deliver what customers now expect personalised, seamless journeys that blend human expertise with AI-powered efficiency. And with only 8% of AI projects succeeding today, the retailers who get it right will stand out where it matters the most; in the eyes of their customers.
Sign up for our forthcoming webinar to learn how you could build a pragmatic AI strategy that connects today’s operations with tomorrow’s opportunities.
