Friend 2.0 is more than a hardware refresh: it is a bet that consumers will pay a premium for an AI presence, even when the product’s practical utility is difficult to define and the category’s most visible competitors have already struggled.
Friend is raising the price of AI companionship.
The company founded by Avi Schiffmann has introduced Friend 2.0, a new version of its necklace-like wearable that adds a built-in speaker and a persistent voice personality. The device costs $249, compared with $99 for the original version launched two years ago. That increase is not a minor adjustment to reflect improved components. It is a commercial test of whether an AI product can command more money when its central promise is emotional rather than functional.
Friend is not presenting the device primarily as an assistant that retrieves information, manages calendars or replaces smartphone tasks. Its promotional material instead depicts people discussing personal subjects, relationships and creative ambitions with the wearable. Schiffmann has described Friend as a confidant—neither an assistant nor a lover—placing the product in a still-uncertain space between consumer electronics, software subscription and simulated relationship.
That positioning gives Friend 2.0 significance beyond its hardware specifications. The speaker may be a straightforward technical addition, but it changes the nature of the product’s interaction with its user. A text-based chatbot is something people open. A voice-enabled necklace can appear to be present continuously. That distinction could improve engagement and make the product feel more intimate, but it also brings privacy, dependency and manipulation concerns closer to the center of the business model.
The strategic question is therefore not whether Friend has made a better necklace. It is whether the company can build a durable market for devices that sell access to an always-available artificial companion—and whether customers believe that relationship is worth $249.
From AI utility to AI presence
The first wave of high-profile AI hardware was largely sold as a way to make artificial intelligence more convenient. These products promised to move beyond the smartphone by allowing users to ask questions, capture information, search the web or perform tasks through voice and natural interaction.
Friend is taking a different route. Its value proposition is not clearly defined by the number of tasks it can complete. The company is emphasizing presence: a device that participates in a user’s emotional life and remains available for conversation.
That difference matters because utility and companionship are measured differently. A conventional assistant can be evaluated by whether it saves time, provides accurate information or completes a task. An emotional companion is judged by less stable criteria: whether it feels attentive, whether its personality is appealing, whether conversations are satisfying and whether users develop a habit of returning to it.
For a hardware company, that can be both an opportunity and a risk.
The opportunity is that emotional engagement may create stronger retention than a set of occasional utility features. If users regard Friend as a continuing presence rather than a tool, they may wear it more frequently and form a routine around it. A product that becomes part of a person’s daily emotional rhythm could have more durable usage than a device purchased for a few novelty demonstrations.
The risk is that the company must deliver an experience that feels consistently personal without overpromising what the technology can provide. A companion that is repetitive, emotionally tone-deaf or unreliable may not merely fail to complete a task. It may undermine the relationship the product is trying to establish.
That creates an unusual product challenge. Friend must make the AI feel sufficiently individual to justify the category, while ensuring that the personality remains broad enough to work for many customers. It must respond quickly enough to feel present, but not so intrusively that the device becomes exhausting. And it must encourage frequent interaction without creating concerns that the business is optimizing for dependence rather than user benefit.
The $249 price makes those requirements more demanding. At $99, an AI wearable can be framed as an affordable experiment or novelty purchase. At $249, customers are more likely to compare it with established consumer electronics and ask what they are receiving in return. The device must compete not only with other wearables, but with smartphones, earbuds, messaging services and increasingly capable AI applications that can already provide conversational experiences at lower or no additional hardware cost.
The speaker is a small feature with large commercial consequences
The most visible change in Friend 2.0 is the addition of a built-in speaker. Yet the strategic importance of that feature is much greater than its engineering complexity.
A voice changes the interaction from a private exchange with a screen into something closer to an ambient conversation. It allows the company to give Friend a more embodied identity. The device can speak to the user while remaining around the neck, rather than requiring them to look at a display or read a response.
This helps solve a central problem for AI companions: presence is difficult to convey through text alone. Text messages can be frequent and personal, but they still require an active decision to open and respond. Voice reduces that friction. It can make the assistant seem more immediate and can turn short exchanges into part of the user’s surroundings.
That may improve engagement, which is crucial for any AI product whose value depends on repeated interaction. The more often a user talks with Friend, the more likely the device is to become part of a routine. A routine can improve retention, generate behavioral data and create opportunities for future services.
But voice also raises the stakes of every interaction. A written response can be skimmed or ignored. A spoken voice occupies physical space and can affect the atmosphere around the user. It may be heard by other people, reveal private information or create social friction in public. The product’s form factor makes those issues more consequential because it is designed to be worn rather than used in a controlled session.
The speaker also reinforces the company’s claim that Friend is a companion, not merely a software interface. That distinction is important for marketing, but it may increase regulatory and reputational exposure. Products that encourage emotional attachment are judged not only by whether they function, but by how they influence behavior. Questions about disclosure, consent, data handling and boundaries become central to product design.
For Friend, the speaker therefore does two things at once. It makes the product more compelling by giving the AI a voice and a stronger physical presence. It also makes the company more responsible for the emotional and social consequences of that presence.
A difficult comparison with AI hardware’s first wave
Friend 2.0 arrives in a market where ambitious AI wearables have not yet demonstrated mainstream demand. Humane’s AI Pin is the clearest warning. The device was positioned as a new interface for computing and AI, but it shut down in less than a year after weak sales.
The comparison is not exact. Humane focused on replacing or reducing reliance on a smartphone, while Friend is presenting itself as a companion. Still, the commercial lesson is relevant: a compelling narrative about the future of hardware does not automatically produce repeatable consumer demand.
AI wearables face several structural challenges. They must justify a separate device when users already own smartphones and wireless earbuds. They must provide enough value to overcome the inconvenience of charging, wearing and learning another product. And they must perform reliably in real-world environments, where connectivity, noise, privacy and social context can undermine demonstrations.
Friend’s decision to avoid a direct smartphone-replacement pitch may be strategically sensible. It does not need to convince consumers that a necklace is a better general-purpose computer. Instead, it can argue that the device offers an experience existing hardware does not: a persistent artificial relationship.
That narrower positioning could make the product easier to explain. It could also make the addressable market smaller. Many consumers may be interested in conversational AI, but far fewer may want to wear a dedicated object for that purpose, particularly if the same basic interaction is available through a phone.
The company’s challenge is to show that the wearable form creates a meaningful advantage. If the speaker merely allows users to hear the same kind of responses they could receive from a chatbot, the hardware may look unnecessary. The product must make continuous availability valuable enough to justify the device and the price.
This is where Friend’s business model will be tested. The company is not only selling a piece of electronics. It is selling a reason to keep that electronics product close to the body throughout the day.
The price increase changes the customer proposition
Moving from $99 to $249 places Friend 2.0 in a more demanding part of the consumer market. The difference is not just $150. It changes the type of purchase the customer is making.
At $99, buyers can treat the product as an experiment. The price is low enough to support curiosity, gifts and impulse purchases. Some users may accept limitations because the financial risk is modest.
At $249, the buyer is more likely to ask whether the product will remain useful after the initial novelty fades. The device must compete with other discretionary spending and with technologies that already have proven functions. It must also overcome the possibility that AI companion behavior can be replicated by an application running on hardware the customer already owns.
Friend’s pricing suggests that the company believes the emotional experience itself has become more valuable. The speaker and persistent voice personality are being treated as features capable of supporting a premium. That is a significant assumption. Hardware businesses usually justify higher prices through better materials, improved performance, broader functionality or a recognizable ecosystem. Friend is relying more heavily on the perceived quality of the relationship.
That may be a viable strategy if the company can create strong attachment and retention. Subscription businesses routinely charge for access to services whose value is experiential rather than physical. But the economics of AI companionship are not automatically attractive. Conversational systems require ongoing computing and infrastructure costs, while the product’s revenue may be concentrated at the initial hardware sale unless Friend also charges for software access.
The supplied launch information does not establish the full structure of Friend’s ongoing revenue model. That uncertainty is strategically important. A $249 device can generate upfront cash, but a companion requires continuing service quality. Personality development, model improvements, safety systems and customer support all create recurring expenses. If Friend needs high-frequency usage to prove the product’s value, it may also incur higher inference costs as engagement rises.
That creates a tension at the center of the business. The company wants users to interact frequently because habitual use is evidence of product-market fit. But frequent conversation can be expensive to provide. Friend must eventually demonstrate that customer lifetime value exceeds the cost of hardware, AI infrastructure, distribution, support and acquisition.
The price increase could help fund that effort, but it also raises the barrier to adoption. A smaller user base may be more engaged, yet limited scale can make it harder to spread the product socially or generate enough data to improve the experience. A lower price could maximize experimentation, while a higher price could support better economics per customer. Friend is choosing the latter path before the category has proved that demand is broad.
The billboard backlash is part of the brand—and a warning
Friend has already become a cultural flashpoint. Its New York City subway billboard campaign went viral after critics repeatedly defaced advertisements that appeared to promote replacing human connection with algorithmic companionship.
The backlash is not a side issue for the company. It reveals how the public interprets the category. Consumers may accept AI for search, drafting or customer service while reacting much more strongly to products that ask them to treat software as a friend.
That distinction creates a complicated marketing dynamic. Controversy can generate awareness at a cost that a young company could not otherwise afford. A provocative campaign can make Friend recognizable, encourage discussion and turn criticism into evidence that the product is addressing an important cultural question.
But attention is not the same as demand. Viral criticism may increase brand familiarity without increasing willingness to buy. It can also define the company in the minds of potential customers before the product has had an opportunity to establish its own value. If Friend becomes known primarily as the company trying to replace human relationships, its intended framing as a confidant may be difficult to sustain.
The company’s messaging will therefore need to balance ambition and restraint. It must communicate why an AI companion is useful without implying that artificial interaction is an adequate substitute for family, friends or community. It must make the personality feel engaging without suggesting that the system possesses human understanding. And it must explain the product’s boundaries clearly enough that customers do not mistake simulated responsiveness for genuine emotional reciprocity.
The billboard reaction also points to a broader competitive issue. Friend is not competing only against other AI products. It is competing against social norms. A wearable that talks to its owner in public may be viewed as unusual or isolating, particularly if the device visibly signals that the user is engaged with an artificial companion. Adoption may depend on whether the product becomes socially legible—something people can wear without having to explain—or remains a curiosity associated with loneliness and technological overreach.
Privacy is inseparable from the companion model
An always-available voice device creates privacy questions that are more serious than those facing a chatbot used occasionally on a screen.
The product’s appeal depends on continuity. Friend is meant to participate in personal conversations and remain available across different contexts. That makes the handling of audio, transcripts, personal history and inferred preferences central to the customer relationship.
A companion that remembers personal details may feel more valuable. It can respond with greater continuity and appear more attentive. But memory also increases the sensitivity of the data involved. Conversations about relationships, emotional struggles or creative ambitions may reveal information that users would not want exposed, repurposed or retained indefinitely.
The wearable form adds another layer. If the device is operating in the presence of other people, those individuals may be captured in conversations without expecting to interact with an AI system. The speaker can also reveal information to bystanders. A response about a private subject may become public simply because the user is on a train, in an office or walking with friends.
These issues are not merely legal or ethical considerations. They can affect adoption and retention directly. Customers who are uncertain about who can access their conversations may reduce usage, disable features or abandon the product. For a companion service, mistrust is especially damaging because the product’s value depends on disclosure and repeated personal interaction.
Friend will need to make privacy controls understandable rather than burying them in technical documentation. Users should be able to know when the device is listening, what is stored, how memory works and how to remove personal information. The company also needs to explain how it prevents the product from becoming manipulative, particularly when its commercial success depends on sustained engagement.
The more Friend emphasizes emotional presence, the more customers may expect the company to behave like a steward of a relationship rather than simply a hardware vendor. That raises the standard for transparency.
The real moat would be the relationship, not the necklace
From a competitive standpoint, the physical device is unlikely to be Friend’s strongest long-term defense. A speaker can be added to other wearables. Voice interfaces are becoming common. Large AI companies and consumer electronics manufacturers have the resources to integrate conversational systems into earbuds, glasses, watches and phones.
Friend’s potential moat would instead come from the software experience and the accumulated relationship between the user and the AI. A distinct personality, reliable memory, strong conversational quality and an emotionally coherent product identity could make switching less attractive.
That kind of moat is difficult to build. It depends on more than model capability. The system must maintain an appealing tone across thousands of interactions, recognize when not to speak, handle sensitive subjects responsibly and preserve continuity without becoming intrusive. It also has to give users a reason to prefer Friend’s personality over a general-purpose assistant from a much larger company.
The competitive pressure will intensify if AI platforms decide that companionship is a valuable use case. Companies with established models, distribution and cloud infrastructure could offer similar experiences through devices customers already own. Their advantage would be lower hardware friction and potentially lower prices. Friend’s counterargument must be that a dedicated object creates a level of intimacy and commitment that a general-purpose platform cannot reproduce.
That is a credible product hypothesis, but not yet a proven business advantage. Dedicated hardware can strengthen a brand if the form factor is essential. It can also become a liability if customers conclude that the software—not the necklace—is the real product.
Friend’s execution will determine which outcome prevails. The company must deliver hardware that is comfortable, dependable and discreet while building a personality that users recognize as distinctive. It must turn the novelty of speaking to a necklace into a durable habit. And it must demonstrate that the relationship produces enough value to survive comparison with free or cheaper alternatives.
What success would look like
For Friend 2.0, success should not be measured only by units sold at launch. A large initial response could reflect curiosity, publicity or the unusual nature of the product. The more meaningful indicators would be repeat usage, retention, customer referrals and the proportion of buyers who continue wearing the device after the first several weeks.
The company will also need to show that its users are not merely testing the companion but incorporating it into daily routines. That would validate the decision to add a voice and charge a premium. If people wear Friend frequently, initiate conversations regularly and describe the device as useful in specific recurring situations, the company may have evidence of a new category rather than another AI hardware experiment.
The opposite outcome would be equally instructive. If customers purchase the device but stop using it, the speaker will have functioned as a novelty rather than a durable interface. If users prefer interacting with the service through existing devices, Friend may discover that the emotional experience does not require dedicated hardware. And if privacy concerns prevent people from using the product openly, the wearable form could limit rather than expand engagement.
The company’s public positioning will matter as much as its technology. Friend is asking consumers to accept a new social role for an AI system. It cannot rely on technical improvements alone to overcome skepticism. It must explain what the companion is for, what it is not, how it treats personal information and why wearing it provides value that a phone does not.
A high-stakes test of AI’s next consumer market
Friend 2.0 represents a broader shift in the AI market from selling automation to selling attachment. That shift could open a large new category if consumers value artificial companionship as a persistent service. It could also expose the limits of AI products that are designed around emotional novelty without a clear economic or social role.
The $249 price makes the test unusually clear. Friend is not asking users to make a small bet on an experimental gadget. It is asking them to pay a premium for a device whose primary product is an ongoing relationship with software.
That strategy has advantages. Emotional engagement could create stronger retention than task-based utility. A dedicated wearable could give the AI a sense of continuity that applications lack. And a focused identity may help Friend stand apart from general-purpose assistants competing on breadth.
But the liabilities are substantial. The category has already been associated with weak hardware demand. The product’s utility is difficult to evaluate using conventional metrics. Privacy risks increase as the AI becomes more personal and continuously available. Public backlash can generate attention while making adoption socially uncomfortable. Larger competitors can potentially reproduce the software experience on existing devices.
Friend’s central competitive advantage, if it develops one, will not be the built-in speaker. It will be the company’s ability to make a simulated relationship feel valuable without making it feel deceptive, invasive or disposable. That requires disciplined product design, sustainable economics and clear boundaries around what the system can and cannot be.
The launch is therefore a business experiment disguised as a hardware refresh. Friend is testing whether companionship can support premium pricing, whether voice can turn intermittent chatbot use into a daily habit and whether consumers will accept a device that occupies an emotional space traditionally reserved for people.
If the answer is yes, Friend may help define a new class of consumer technology: hardware built less to perform tasks than to maintain presence. If the answer is no, the company will join a growing list of AI hardware ventures whose narratives were more compelling than their products.
At $249, consumers are being asked to decide whether an artificial friend is worth the cost. The larger question for the market is whether that friend can become a durable business.