“So people are critical in all of this because what this does, this platform and any AI tool for that matter, what it does is it takes really good people and it makes those people have a superpower.”

—Frank Knox, COO of Auto Acquire

AI Is Changing the Used Car Game

CarMax and Carvana didn’t build their edge on flashy ads alone. They’ve been running AI-powered acquisitions for years—automating offers, adjusting them in real time, and casting a net wide enough to dominate consumer-direct inventory.

Frank Knox knows that world inside and out. After nearly two decades at CarMax and later leading acquisitions for a major dealer group, he’s now COO of Auto Acquire, a platform designed to put those same AI tools in the hands of retail dealers.

Here’s what he says AI can do right now:

• Automate and scale offers. Machine learning keeps pricing live, tuned, and competitive—based on your dials, not a generic market feed.

• Reveal the whole garage. With just a name or email, AI can surface every vehicle tied to a household. That means one lead could equal multiple cars.

• Handle the follow-up. Sequenced outreach keeps sellers warm while your people focus on high-intent conversations.

• Streamline logistics. Transport, inspection, ID checks—AI plugs into the friction points that slow deals down.

But Knox is quick to point out: AI isn’t the closer. People are. Managers still tune the dials, validate the condition, and build the trust that makes a seller say yes.

Bottom line: the same tools that once made disruptors untouchable are now dealer-ready. AI doesn’t replace your team—it gives them superpowers.

Guarding Your AI from Poisoned Prompts

It may be time to have a conversation with your AI system and tool providers about how they’re keeping you safe.

Generative AI isn’t just empowering businesses — it’s empowering attackers, too. New research from Columbia University and Barracuda shows that cybercriminals are now using AI to outsmart AI.

Polished, typo-free phishing emails — written by large language models — can slip past spam filters and fool employees. Worse, attackers are learning to hide malicious prompts inside everyday messages. When an AI assistant (like Copilot or GPT integrations) scans that email to summarize or respond, it can execute those hidden commands — leaking data, altering records, or triggering unauthorized actions.

Even AI-powered defenses aren’t immune; poisoned inputs can twist automation tools against you.

Possible Steps

• Isolate AI tools that touch sensitive data (CRM, finance, HR).

• Review permissions — follow “least privilege.”

• Educate teams: AI-written phishing looks perfect, not sloppy.

• Validate outputs: Don’t assume every AI summary is accurate — spot-check before acting.

AI can be your best assistant or your weakest link — it depends on how tightly you manage its access and awareness.

Anthropic just dropped Claude Sonnet 4.5, and it’s rewriting what “autonomous” means. This model ran solo for 30 hours straight, coding an entire chat app — about 11,000 lines of code — with no human intervention.

Why it matters: The leap from 7 to 30 hours puts us squarely in the era of hands-off AI agents. Claude 4.5 can manage multi-step tasks, juggle virtual machines, and coordinate multiple AIs. Anthropic says it now leads the market in “computer use” — executing tasks inside apps, not just advising.

For dealerships, this points to near-future tools that:

• Build full reports overnight, no babysitting

• Run complex spreadsheet analyses or data scrubs

• Manage CRM automations, follow-ups, and reminders across systems

It’s not just a faster chatbot — it’s a self-sustaining worker. The question now isn’t if agents can work — it’s how long you’ll let them run.

Claude Sonnet 4.5 isn’t just smarter — it’s steadier, capable of reasoning over 30 hours and thousands of tokens without losing track. But power like that needs structure. Anthropic’s own engineers stress three keys for great results:

• 🧭 Be explicit — spell out goals, limits, and success criteria.

• 🧩 Add context — tell Claude why each step matters.

• 🧱 Track state — use check-ins, progress notes, or files so it never drifts.

Use this week’s prompt to design your first autonomous project plan — safe, scoped, and dealership-ready.

Prompt example (Claude Sonnet 4.5)

The debut of Tilly Norwood, an AI-generated actress seeking real-world representation, has sparked outrage across Hollywood — not for what she can do, but for what she might replace. Critics warn that synthetic faces built from real ones blur lines of consent, creativity, and authenticity.

For dealerships experimenting with AI voices, avatars, and video hosts, there’s a caution here: audiences crave connection, not perfection. What looks efficient in the boardroom can feel eerie or deceptive to couch-shoppers.

In an age of authenticity, the wrong AI presence could undermine trust faster than a bad review. Customers don’t want to feel “sold to” by a script — they want someone real guiding the experience.

Takeaways:

• Use AI to amplify human connection, not replace it.

• Disclose when AI is speaking.

• Pair virtual tools with real names, real faces, and genuine follow-up.

In retail, as in entertainment, it’s not just about who’s on screen — it’s about who your audience believes they’re meeting.

Ford CEO Jim Farley is sounding the alarm: America’s AI ambitions will stall without rebuilding the essential economy — the skilled trades that power factories, supply chains, and yes, dealerships.

Farley’s warning hits home for auto retail: even as digital tools transform sales and marketing, service bays are running short on technicians, and blue-collar productivity is falling behind AI-powered white-collar sectors. The gap isn’t just talent — it’s culture. Families still steer kids toward coding, not wrench-turning.

That means two things:

1. Talent strategy is now an innovation strategy. Invest in apprenticeship programs, partner with local trade schools, and market technician careers as high-tech, high-impact work.

2. AI belongs in the bay. Tools like AR repair guidance, predictive diagnostics, and smart scheduling can help smaller teams do more with less.

Farley’s bottom line: you can’t “re-shore” or rebuild without people ready to do the work. In automotive, that starts in your own service drive.