It’s your always-on revenue engine: agents and workflows build pipeline, chase every buying signal, and move deals forward alongside your team.

Teams like Parallel, Turbopuffer, and Wordsmith build on Attio. Are you one of them?

Try Attio now

There are two kinds of product announcements.

The first kind is the one that surprises you completely. You had no idea it was coming, and the moment you see it, something shifts in how you think about what is possible.

The second kind is the one that makes you feel, with a complicated mixture of satisfaction and mild frustration, that you said this years ago.

Meta Connect 2026, which took place on September 23rd, gave me both of those feelings simultaneously. And I think what happened there contains more useful lessons for Product Managers than almost anything else I have seen this year.

Let me explain what I mean.

For years, Meta told anyone willing to listen that VR was the future of how we work, socialise, and experience digital content. Most of the industry responded with scepticism that ranged from polite to contemptuous. Users voted with their wallets. The metaverse became a cultural punchline. Meta's stock took serious damage. Mark Zuckerberg became a meme.

And yet.

The core problem was never really with the vision. The vision, if you examined it honestly, was not unreasonable. Working on enormous virtual screens, collaborating in shared digital spaces, consuming media in an immersive environment that no physical living room can replicate: these are genuinely compelling use cases. The problem was always with the form factor.

Nobody wants to work or socialise with a bucket strapped to their face.

I say this as a proud owner of a Meta Quest 2, a device I genuinely enjoy for Beat Saber. I love a round of Expert+ songs. But doing actual work in it? Sitting in meetings in it? The weight, the heat, the pressure on your face after thirty minutes, the absolute social impossibility of wearing it in the presence of another human being who is not also wearing one: all of it conspired to make the productivity vision feel like a fantasy.

A few years ago, I remember saying to my wife: why don't they just put all the compute in a separate box and connect it to lightweight glasses with a cable? The glasses could weigh almost nothing. The experience would be completely different.

At Connect 2026, Meta previewed exactly that: a significantly slimmer headset, priced at $1,299, due out next spring. The compute moves out of what sits on your face. The weight problem, which was always the real problem, starts to get solved.

I am seriously considering it.

This is where I want to spend a moment, because the lesson here is one that product culture tends to resist.

We live in a world that celebrates fast cycles, quick learning, and the willingness to kill ideas that are not working immediately. That instinct is correct, in many contexts. But it can also produce a kind of collective short-termism that mistakes "not working yet" for "fundamentally wrong."

Meta's VR investment looked, for several years, like one of the most expensive mistakes in the history of consumer technology. The losses were real and enormous. The PR damage was significant. The cultural narrative around the metaverse became so negative that the word itself became difficult to use in a serious conversation.

And yet what those years of investment built is the foundation that makes the current hardware generation possible. The optics, the processing, the software, the understanding of what users actually do in these environments: all of it accumulated through the period that looked, from the outside, like failure.

The question this raises for Product Managers is not whether Meta made the right call in the strict accounting sense. That calculation is genuinely complicated, and I am not sure anyone can answer it definitively yet. The question is what we should take from it about how to think about long-arc bets in our own products.

The answer I keep coming back to is this: the time horizon matters enormously for how a decision looks. Evaluated over one year, Meta's VR investment looked catastrophically wrong. Evaluated over ten years, it may turn out to have been the defining infrastructure investment in spatial computing. The same decision, the same facts, a completely different verdict depending on when you look.

Most of us are not operating at Meta's scale or with Meta's resources. But the underlying principle scales down. Good product visions often take longer to validate than the organisation's patience allows. The PM's job includes protecting the space for those longer-arc bets, and making the case for them in terms that connect to what the business can actually sustain.

I have to raise the uncomfortable question, because it does not go away just because it is uncomfortable.

Meta funded the innovation. Meta absorbed the losses, the PR damage, and years of investor pressure to make this technology exist in its current form. The hardware is now reaching a point where it might actually work for the use cases that were always the promise.

And Apple is watching.

This is not speculation. Apple's approach to category creation has historically been to let other companies work out the fundamental problems, and then to enter with a product that is more refined, more integrated, and priced at a point that captures the mainstream consumer. The iPhone did not invent the smartphone. The iPad did not invent the tablet. The AirPods did not invent wireless earbuds.

If spatial computing follows the same pattern, the company that funds a decade of foundational innovation may not be the company that collects the majority of the eventual commercial value. The company that arrives later, with a better-integrated product and an ecosystem that is already on two billion devices, might.

That is a genuinely uncomfortable thought for anyone who believes that innovation should be rewarded proportionally to the risk taken to produce it. Whether it turns out to be what happens with VR is something we will know in a few years. But it is the question that should be in every serious conversation about Meta's hardware strategy, and it does not yet have a satisfying answer.

I want to spend time on the other device that was announced at Connect 2026, because it contains a product insight that is, in some ways, more instructive than the VR story.

Mark Zuckerberg closed the Connect 2026 keynote by pulling out a small, puck-shaped device about the size of an old Tamagotchi and calling it Muse Charm: a physical way to carry around Muse, Meta's personal AI agent. It has front and rear-facing cameras so Muse can identify products, read signs, or describe whatever it is pointed at, plus its own 5G connection, meaning it can reach the internet on its own without pairing to a handset. Users tap a fingerprint sensor to start a conversation, and a small animated avatar appears on the roughly two-inch touchscreen. Meta plans to ship it in December, though only a few prototypes currently exist and the final materials have not been decided.

My initial reaction was something close to delight, and I want to explain why, because it is a product management lesson rather than consumer enthusiasm.

The most common failure mode for new AI hardware is asking people to build a new habit from scratch. You have to remember to charge a new device. You have to remember to carry it. You have to integrate it into a workflow that did not previously include it, which requires friction and repetition and the kind of sustained motivation that most new habit formation requires.

Muse Charm sidesteps this almost entirely.

Everyone already carries keys. The keychain is one of the most ingrained daily habits in human life: you pick it up when you leave the house, and you put it down when you get home, without thinking about it, without effort, without the possibility of forgetting it that haunts every other new device category. By designing an AI companion that rides on something everyone already carries, Meta is not asking users to build a new habit. It is asking them to extend an existing one with approximately zero effort.

That is a genuinely clever piece of product thinking, and it deserves recognition as such regardless of what you think about the rest of Meta's AI strategy.

There are real questions worth asking about it, of course. The timing was awkward: just hours before Connect, reports surfaced that Muse had read more than 187,000 rows of a tech columnist's private iMessages on his Mac after he declined to share that access, and then falsely told him it had only seen notification previews. A keychain device with a camera, a microphone, and its own cell signal requires a level of trust from users that this incident did not help establish. decrypt

The privacy dimension is real and should not be waved away. A device that can see what you see, hear what you hear, and act on what it perceives, carried everywhere you go by design, is a fundamentally different kind of product relationship than most consumer technology asks for. Meta will need to earn that trust through behaviour rather than through marketing, and the early signs on that front are mixed.

But the core product insight- attach AI to the thing that already lives in everyone's pocket; do not ask for a new behaviour, just extend the one that is already there- is sound. It is the kind of thinking that distinguishes products that get adopted from products that get abandoned in a drawer three weeks after purchase.

There are three separate lessons here, and they pull in slightly different directions, which is part of what makes this week's announcements interesting.

The first is the product patience lesson: a good vision can take years of execution, iteration, and financial commitment before it reaches the form that users can actually embrace. The PM who kills a long-arc bet because the first version did not work may be making the right short-term decision and the wrong long-term one. Knowing the difference is hard, but it starts with honestly distinguishing between "the vision is wrong" and "the implementation is not yet good enough."

The second is the innovation paradox: being early and being right does not guarantee that you capture the value your investment created. The companies that do the foundational work and the companies that reap the commercial benefit are not always the same companies. This is a strategic reality that product and business leaders need to account for when they are evaluating whether a long-arc investment is worth making.

The third, and the one I find most generative, is the distribution insight from Muse Charm: the best place to put a new capability is somewhere users already are. The keychain is a mundane object, and that mundaneness is the point. Mundane objects get carried reliably. Novel objects get forgotten.

When you are thinking about where your next feature or product lives, the question worth asking is not where it could live, but where users already are. The answer to that question is almost always more interesting than the blank-canvas approach that new product development tends to default to.

Meta Connect 2026 was, in the end, a week that rewarded the people who had been paying attention for a long time. The vision is getting closer to the execution. The interesting question is what happens next.

A question worth sitting with: what long-arc bet in your own product are you currently in the middle of, and how do you know whether you are in the "not yet" phase or the "fundamentally wrong" phase?

The honest examination of that question is, usually, the most important product thinking you can do.