Apple’s recent announcement of so-called ‘Apple Intelligence’ at WWDC raised more than a few eyebrows. Some of this may relate to the extent of the privacy Apple has been able to offer, while much will be directed at the deep integration with OpenAI. Regardless, there is an open debate about how safe such an integration might be, especially given that OpenAI is no stranger to controversy.

But what can Apple’s approach to AI enablement teach us about integrations in Mews?

Choose partners wisely

OpenAI is not only one of the first major players in the generative AI space, but they are also one of the most controversial. Whether it’s the embattled CEO Sam Altman making headlines, their approach to intellectual property, or the misappropriation of an artist’s likeness, the company gets as much negative press as it does positive. Apple’s decision to partner with OpenAI has therefore raised eyebrows, given the former’s historic championing of privacy concerns. This is the company, after all, that put a leash on Facebook.

Questions a company might ask itself when deciding to integrate with a partner:

• Would this news be celebrated or deemed controversial?
• Is this partner’s positioning and reputation in line with our company’s identity?
• Would discussion focus on the merits of this integration, or something else?

Even if there were potential controversy to be caused, it might be possible to mitigate—but the first step to doing this would be awareness of any issues.

For example, it remains to be seen whether deep integration at the OS-level such as this is simply opening Apple up to yet another antitrust case due to undermining consumer choice

Communicate clearly

One of the most interesting details about the broad discourse surrounding Apple Intelligence is that it is not inherently insecure. The implementation Apple has chosen means that all use of AI* is performed either on-body, or in a private cloud service, such that even Apple cannot access users’ data. Only if the user explicitly invokes the OpenAI service will data be shared outside this effectively secure enclave. This model clearly puts security first—but the amount of public concern that has been voiced would imply this message simply hasn’t gotten through. Although, in fairness, the vast majority of this concern appears to have been voiced by one man.

What this goes to demonstrate is that even a service offering which has strong customer value and aligns with user interests may be misunderstood by the market if the vendor is not careful. It is therefore vital to speak to customer interests and be totally explicit when allaying concerns.

This is doubly true for business-to-business offerings, where the problem space is inherently more complex, and the resulting technical literacy needs for customers are commensurately higher.

One area for potential misunderstanding is when an integration is required to ensure a property using Mews is compliant with legal requirements. How a particular integration works, what responsibility the customer has for configuring and using it, and any remaining gaps in compliance all need to be communicated clearly. It can also be challenging to understand, when issues occur, with which party the blame may lie.

Keep the pricing simple

Another detail which has only just come to light is that Apple may be paying OpenAI in ‘exposure’ (ironically a Gen AI platform which ‘replaced artists’ has suffered the same fate they did). However, while this may be reporting conjecture, what is known is that Apple will not be charging its customers to use OpenAI. It’s fair to say that pricing doesn’t get much simpler than this!

When integrating with a partner, many pricing models are available. Depending on the vendor relationship, the integrating party may contract with Mews, or with its customers directly. If customers set up their own accounts, then they would individually pay for usage. At lower rates of use, this is likely a more expensive way to access services.

However, if Mews is the contracting party, an economy of scale might prompt lower unit costs, such as for invoicing services. This leaves Mews with the choice of whether to pass these costs through, write them off as the price of market entry, or factor them into market pricing.

This decision is a sensitive one and should factor in all other aspects of market presence, but critically, it should be easily understandable by customers. The selling and onboarding processes depend on clarity, and for Mews to be evaluated alongside its competitors, it must be possible to accurately project costs. A nasty surprise during system setup does not benefit lifetime value (LTV); but is it worth elevating customer acquisition costs (CAC) to avoid this?

A civil partnership

There are many factors that go into decisions about when to partner, whom to partner with, how to design the partnership and how to price and communicate this to customers.

What other concerns should businesses address when considering this? How much room for negotiation should there be? And how and when should such partnerships be re-evaluated? Share your thoughts!