How these four AI brands are trying to solve their differentiation problems

One of the most fascinating parts of being a strategist is that you eventually stop thinking in terms of mutually exclusive right and wrong answers, and you start entertaining the notion that things may be complementary, and trade-offs are all we ever have.

This is certainly something I took away from the conversation I had with Julian Cole, where we looked at four recent brand campaigns – by Anthropic, OpenAI, Harvey and Legora – and tried to dissect what was working or not. A fun exercise, done with good intention.

Of course, this is only the opinion of two outsiders who had nothing to do with the work, relying on publicly available information, experience and judgement to unpack work that, one way or another, reinvigorated why we enjoy this craft in the first place.

You can watch the full conversation above, or below are some condensed reflections I wrote down on each of these brand campaigns. This isn’t a definitive take on each of them, not by a long shot, but rather an invitation to have more of us come together and have a thoughtful discussion about work we’re interested in.

When everyone’s too worried about whether something is genius or horrendous, and there’s very little in-between in the discourse, some more systematic analysis might be what helps us reduce cognitive atrophy due to said AI platforms.

Works well enough for me, anyway. So let’s get started.

Claude: The problem solving brand

One of the lenses through which I look at Anthropic’s work around Claude is that it’s very clearly not trying to be ChatGPT. Which is to say, it’s not trying to say it’s the tool for all use cases, or at least that’s not the main focus of their comms. It’s much more attitudinally aligned around who they’re for: problem solvers.

Why this works

Because they acknowledged that deep work, high agency and cognitive extension are preoccupations among many of us, and this is especially true among knowledge workers. Well, the fact Claude is winning on B2B adoption means they want to keep cementing that role by being associated with superior problem solving skills inside an organisation, rather than productivity slop.

Where it may fall short

While finding a sub-segment within the entire category is a smart move when you’re the challenger brand, they cannot just stop at identifying the segment they’re for. They need to keep building consistent associations that make sense for a B2B audience, and also explore what problem solving might mean for a B2C audience if they want to grow penetration there too. Fresh consistency, etc.

Comms ways in

I can’t help but wonder where else they take this platform and communications idea next. They’ve already done the Super Bowl, showed up in OOH, started communicating their sponsorship with the Formula One (gorgeous art direction in some of those films btw). It’s loads to work with and build on.

One area I suspect we’ll see more of Claude in action is through creator collaborations, partnerships with makers, unexpected demos on the street or at events. All in order to continue reinforcing the fact that they are for problem solvers of all kinds, whether you’re enterprise or not.

ChatGPT: The utility brand

ChatGPT has a tremendous business problem which we might call ‘shallow usage’. Loads of people use them, but only every once in a while. This means that, while on MAU measures they might be seen as a mass market product, this doesn’t mean that usage translates necessarily into value the business can capture through paid subscriptions (why pay for something you only use sporadically?). Hence, their focus on growing an advertising business on top.

So, when they launched the platform around the everyday magic of AI, it makes total sense that they want to position themselves as an everyday utility in order to grow overall usage occasions. It’s just an utility, like electricity, which works largely in the background.

Why this works

They’re the market leader by B2C usage (even if shallow) and definitely have a more famous brand than Claude, albeit not necessarily a stronger brand if we’re looking at brand being a lever to grow pricing power. However, by being the leader, it makes sense that they want to be equivalent to AI usage as a whole, because if the category grows they benefit from that. The brief might have been around showing frequency of usage, not necessarily showing the benefits of trying it for the first time.

Where it may fall short

While showing emotionally warm advertising around the different usage occasions of an AI tool is important, it may not be sufficient to offset the fact that there’s a lot of reputational damage that needs fixing. They need to sustain their investment and hope that the reputational damage at a business level doesn’t accelerate beyond what brand building communications can compensate for.

Comms ways in

Because they’re going for variety of usage occasions, and have a significant reputational problem to overcome, they need to keep investing in PR, partnerships and creator partnerships. On that last one, I’d specifically invest in specific interest areas that have nothing to do with tech, so you might see more cooking creators talk about inventing recipes with ChatGPT or medical professionals talking about how it helps them make diagnostics more efficiently and cross-check sources more effectively.

Legora: The sexy brand

Ah yes. The joke we all probably made at some point. Jude Law should feature in an ad about lawyers! And look, I gotta tell ya, the tone of this kinda worked for me because I have a very deadpan sense of humour. But there’s only so much a famous face can do for the brand and product, as we’ll see in a short minute.

What works

I’m not a lawyer but I’d bet lots of people who are lawyers love the idea of being a lawyer, even if not always the reality. There’s a sense of powerful ego, status games, identity plays, and career compounding effects that lawyers aspire to. In this sense, saying that law just got more attractive bypasses any rational objections to using AI and goes straight into making the job feel aspirational. Indirectly, AI makes the grunt work easier, but the point is not that, it’s what happens after that. You get to live your sexy lawyer life.

Where it may fall short

I really want this to be a long-term partnership, just like George Clooney and Nespresso (and now Dua Lipa!) are a long-term partnership, because otherwise it will feel like a one-off joke that turned out to be very expensive because celebrity fees. So the building block of making law feel more attractive is interesting, but it needs to be sustained in order to build actual meaning into it.

The other part of this, which Julian pointed out, is that actually what we might take away from the comms is not the brand name, but the category association: law. And this may be a challenge, especially because Legora is not #1 in the category (Harvey is), which means the work needs to work harder to associate their positioning (make law sexy again) with their celebrity endorsement (Jude Law) but in order to make the brand name famous (Legora).

And right now, I’m not sure it’s doing that. I’d be curious to see if there was System1 or Tracksuit or Behavio research on any of this.

Comms ways in

There’s a potentially interesting, high end, status games world here that Legora could occupy in the long run. If they adopt the body language of a fashion or luxury brand, or a jet setting celebrity that does high concept collaborations, shows up in high stature buys (big ass posters basically, takeovers) and high end glamour occasions (celebs need lots of lawyers!), they might build something distinctive on top of the ingredients they do have.

Harvey: The corporate shield

Harvey is a slightly different shade of situation from Legora. They’re the leading brand in the legal AI category, and therefore are focused on building trusted systems and institutional safety that customers can rely on in their biggest time of need. The last thing you want is for an AI to hallucinate evidence while you’re preparing for a court appeal. Alas, none of this is clear in comms, but rather that they hired the guy who played Harvey Specter.

Why it works

There is a part of me that wants to give the benefit of the doubt to Harvey for hiring the actor who played the character who gave the company its name, but even just reading that sentence is tiring and that partly explains the point. It’s a long leap. Now, there isn’t anything inherently wrong with finding a celebrity that frankly only you could hire for this particular reason, and that element of pure distinctiveness may do something to help build a healthy brand.

Where it may fall short

But then we realise that, up until this point, all we have is a clever joke about a TV character that not everyone may understand. It may feel funny for those who do know the show, but the venn diagram between Suits fans and people who need a lawyer and people who are willing to give AI the benefit of the doubt for their legal affairs might be a bit thin. If they don’t do something more to build the link with Harvey Specter more meaningfully over a period of many years, I worry this will feel like another expensive in joke.

Comms ways in

And yet, there’s always good opportunity when your position is to be the trust custodian of your category, which when you look at Harvey’s website they’re trying to be. The more we see them investing in partnerships, original research and PR-led initiatives, which yes should somehow involve world building around Suits (look, if you’re gonna go for it, really go for it), the more they can make their money’s worth to continue leading the category.

You ready for a quadrant?

Ultimately, all the above is just a bunch of educated guessing by two strategists who wanted to critique some work and try to learn from it. So I hope that, if you worked on any of the campaigns above, you take that in this spirit. And I’d love if folks who worked on this would happily correct any of the above, or show the counter-point, because this is how we all learn to get better.

But I asked you a question about quadrants, and assuming the answer is ‘yes’, and knowing these brands are but a n=4 of the total universe of AI-first brands investing in advertising, here’s a little quadrant on where they fit in terms of their positioning.

	Task-led	Identity-led
Horizontal integration	The Default Layer E.g. ChatGPT Broad capability, anyone, any job. Sells speed and accessibility. Risk: commoditisation.	The Intellectual Bicycle E.g. Claude Sells cognitive partnership to high-agency generalists. Risk: reads as precious.
Vertical integration	The Corporate Shield E.g. Harvey Domain-safe, institutionally adopted. Trust is collective, not individual. Risk: firm buys it, lawyer tolerates it.	The Professional Glow-Up E.g. Legora Sells elite status and career identity to a profession that runs on both. Risk: vanity software if product lags positioning.

Not to say one quadrant is better than the others inherently, but rather that over time we will see which quadrants become more populated by other brands too. And round and round we go again, because all of this is but a cyclical exercise in meaning making.

Check out the full conversation I had with Julian here.