GM's AI showcase put a busy parent up front. Take the cue
If you are launching a product with AI in it, put the job it takes off someone's day in the headline and keep the technology further down the page. General Motors' tech showcase in October gave a useful example, and a separate buyer study offers some evidence on how the wording lands.
A general product, shown through one person's day
At the showcase, GM used a busy parent as one of its examples of how drivers should see their car: as something that takes jobs off their day. The idea frames OnStar AI, a personal assistant at the centre of a relaunch of OnStar, GM's 30-year-old subscription service for connected cars, reports Marketing Brew.
For now, the assistant will be free as part of OnStar's standard connectivity package. It arrives first on the 2027 Cadillac Lyriq, then the 2027 Chevrolet Silverado and GMC Sierra, with more models to follow.
What interests us is the gap between how GM describes the product and the example it chose. Anna Santos, GM's director of product management for voice and AI/ML, told Marketing Brew that planners see OnStar AI as relevant to all GM customers: "It's very general-purpose as a product."
Yet chief executive Mary Barra pictured a father whose car takes his daughter to kindergarten, runs errands and gets him to work, and will one day drive itself through that routine. That was her picture of a self-driving future rather than an OnStar AI advert, but it set the tone.
Michael Wahlstrom, GM's executive director of vehicle product management, said the busy parent is "especially relatable" because families are an important part of GM's customers, adding: "But the opportunity extends beyond parents." He says marketing will focus on OnStar becoming "a more intelligent, proactive part of the ownership experience", tailored to each brand.
Two analysts, Paul Waatti of AutoPacific and Stephanie Brinley of Mobility Global, agreed the parent focus fits the people buying the trucks and SUVs the assistant is rolling out on. Waatti said that is where GM makes a lot of money, and called those families "their easiest customer to convince". That is his reading, not GM's stated reason, but it is the one marketers can use: lead with the customer who buys most.
The evidence: a test of labels
Behavioural scientists at Irrational Labs say in a write-up for Growth Unhinged that they showed 1,014 US software buyers pages for three small-business tools: a calendar, a social media toolkit and accounting software. Only the label changed: none, an AI "agent" (software that carries out tasks for you), or a "team of agents", each doing a different job.
A single "agent" label made interest dip slightly, not significantly. A "team of agents" made people 6.3 percentage points more likely to say they wanted the product than the single "agent", a statistically significant gap; it also beat no label, though not significantly.
The caveats: it measured stated interest, not purchases; what people said they would pay showed no significant differences; it covered three small-business products; the results come from the authors' own write-up; and people expected the "team of agents" version to be harder to use than the no-label one.
The authors advise making each job visible, including only jobs buyers want and keeping setup looking simple, and hope products won't all rebrand as a "team of agents".
What it means for launch teams
Neither story proves the other. GM doesn't call its assistant an "agent", and the test was about software buyers, not people choosing a car. Our own reading of both is that a general-purpose product is easier to grasp through specific jobs for a specific customer.
Marketing Brew reports that GM's pitch also targeted investors, with the message that GM deserves to be valued like a tech company, alongside a planned move from the New York Stock Exchange to Nasdaq. Our reading, not GM's: the technology story is for the markets, the chore story for the customer, and that split is worth copying.
A client may want "AI-powered" at the top of the page because the board likes it; our advice is to keep the technology story for investors and press, lead with the jobs for customers, and test the wording before launch.
For in-house product and brand teams, that means listing the jobs a feature does and leading with the one customers most want gone. The study's authors warn that if one feature feels irrelevant, people may reject the whole package.
Data is the other half of the page. Waatti told Marketing Brew that the more useful OnStar becomes, the more personal the data, and that GM is asking for that trust when it has work to do to rebuild it.
This year the US Federal Trade Commission finalised an order barring GM and OnStar from sharing certain consumer data with consumer reporting agencies for five years. GM says it doesn't use assistant conversations for advertising and lets customers choose what the assistant can access.
What to do this week
- Rewrite the headline as the task it removes. Ask for a version of your AI feature's launch page that leads with the chore, and show both to five customers.
- List the jobs, then cut. Write down each job the feature does and drop any customers haven't asked for.
- Show each job working. Add a short animation of each job to the landing page; the study's authors suggest GIFs. We'd do this before buying paid media.
- Answer the data question early. Put one plain sentence on what happens to customer data near the top of the page, agreed with legal.
Next time you launch something with AI in it, which goes in the headline: the technology, or the job it does? We'd like to hear what your team has tried.
Jules, founder of WYNStrategy Director helping agencies and marketing teams use AI properly. More about Jules