A Look Ahead: What's Next For Frame

It's a good time for me to pause and reflect on my journey with Frame so far and share some of my plans for its future.

I set out to create the easiest way for people to gather and interact in a customizable 3D space, accessible from desktop, mobile, and VR devices right from a link in the web browser. Despite what shortcomings Frame still has and how much easier it could still be, I think I've succeeded. Thousands of people can now join a space in Frame across different devices, simply with a link. Without writing a line of code, users can bring in 3D models, web browsers, videos, images, whiteboards, screen shares, particle systems, shaders, polls, and dozens of other assets. We've shattered dozens of limitations I thought we couldn't shatter. In one moment I'm particularly proud of, the CTO of Microsoft introduced Frame and our AI-infused demo at Microsoft Build last year.

It's become clear to me, though, that none of that is enough. During the pandemic and the peak of remote work, it was easy to convince myself otherwise. Attracting a significant user base was encouraging, but in hindsight, many of those users were seeking temporary relief from Zoom fatigue. They have since returned to the office or, having gotten the relief they needed, didn't find a compelling reason to keep coming back to Frame. We have a healthy amount of free users and growing amount of paid subscribers, but not as many as I'd hoped for. Many of our most passionate users are educators with very constrained budgets.

We have landed a few larger projects that involve us doing custom work on top of the core Frame product: 3D environment design, custom feature development, and branded deployments with dedicated infrastructure. If we focused solely on that line of business, it's possible that it would yield a path to sustainability. We're still a pretty small team, though, and pursuing that path would come at the expense of trying to make an incredible product for the world at large.

Without a shadow of a doubt, I want to make an incredible product for the world at large.

Let me take a step back. I got interested in the broad "metaverse" space because of my research in grad school on teachers and students using virtual worlds for online education. Second Life, anyone? When I discovered Hubs, a brilliant, browser-based metaverse product from Mozilla, it showed me that this sort of social, cross-platform, immersive experience was possible right on the web. With the fire of a thousand burning suns, I furiously learned to code and built the earliest version of Frame - my own take on immersive collaboration. Since then, I've grown the team and I love being able to work with people who are much smarter than me.

For a while, I thought the clearest path to success would be to do what Mozilla Hubs was doing, but to do it a little differently and hopefully a little better in certain key areas. In some respects, maybe we did; in others, we certainly did not. I've enjoyed seeing the Hubs product evolve and I have tremendous respect for everyone who was involved in it over the years. Sadly, news came out earlier this year that Mozilla Hubs is closing down as Mozilla invests elsewhere.

One of my thoughts in response to that news, besides feeling sad about it, was that perhaps some of the Hubs users could find a home at Frame. Heck, maybe we'd get some subscribers out of it, if some of them liked what they saw and wanted to upgrade.

After thinking about it more, I believe that would be a path towards slow, inevitable failure. The lesson to take away from Mozilla Hubs shuttering probably shouldn't be "keep trying to do what Mozilla Hubs was doing, but do it a little differently". I have to be bolder.

There's a lot I have to do. I also need to more narrowly define our target market and bring some real focus to the product. I have to stop bolting on various features that might be compelling in certain contexts, but when combined together result in a confusing, unfocused user interface. I need to dig deeper and establish better product-market fit. I have to re-evaluate our business model.

I don't see any of these as optional. It's all on the table.

So, I've been stepping back from hacking out more features to Frame and I've just been.....thinking! Literally walking around the block alone or getting on calls with others and thinking hard about the best path forward for Frame, doing my best to reflect with clear eyes, an open mind, and an honest assessment of our current situation. Those of you who have been following this journey know that "slowing down" isn't typical for me and the Frame team. Now that I've been doing more of it, though, I can say with confidence that it's something I should have been doing more of over the past few years. Product velocity means little if the direction isn't right.

There's still a lot more thinking for me to do, but I've arrived at the outlines of a plan. We've already begun executing on it. As I've been hinting on social media, I am as fired up as I was when I first started building Frame.

I see tremendous opportunity in front of us to build on the core Frame product to make something focused, extraordinary, and sustainable.

So - What's Next for Frame?

Cue the eye-rolls - artificial intelligence. Hear me out.

You might be thinking, "Wow, Gabe, this is clearly an opportunistic move to latch onto the excitement around the tech trend of the day." In some respects, you'd be right. Look - I'm not above exploring and embracing a trend. You shouldn't be either. That said, I'm also genuinely, unabashedly excited. When ChatGPT first arrived, I wasn't immediately convinced that this was going to be transformative technology.

I am now.

I also have deep reservations about the underlying mechanics of the tech as it currently exists and the way large language models are trained. I hope that the companies training these models find a way to profit from them in a fair way. The fruits of humanity's collective intellectual and creative output shouldn't be exclusively, relentlessly squeezed for profits by the few select companies with enough GPUs to do the squeezing. I expect imperfect solutions.

But the AI-generated cat is out of the bag and it's not going back in. Admittedly, I don't want it to go back in. Software can now do fundamentally different things than it could when I started Frame. I have compelling ideas around the intersection of AI and 3D that need to see the light of day and I'll make sure they do. I'm immensely grateful that I can do this work in the context of Frame and keep building on top of the incredible foundation we've laid so far.

What do I mean by artificial intelligence in Frame? What has me so excited?

I'll start by telling you what it won't be. It won't be what you've seen to date in Frame and many other metaverse platforms: a few scattered features bolted in that might be cool but which lack cohesion and focused purpose. When I first added AI image generation and skybox generation to Frame, my primary reason was so that I could basically say "Look! We do AI too!" That's not a strategy and that's not a forward-looking product vision. It's closer to desperation.

So, what will it be? I'm happy to share.

We have a phased plan that starts with a goal that might sound obvious or already played out: supercharging online meetings with generative AI. The truth is, I don't think anyone is nailing this yet, and if they are, I don't think they are going to do it as well as we're about to. There are a few angles to the value proposition here.

For starters, millions of people are using generative AI to create things, but very few tools help them do it together in real-time. We're going to make Frame the best way for people to come together and prompt, create, and explore together. Here's a video that gives a glimpse of some of the things our creative AI agent can already do.

We'll be helping people do that together, in a user interface that facilitates the creation of and discussion about a wide range of different kinds of content. Prompt together. Ideate. Brainstorm.

Another crucial angle, though, is around discussion and dialogue. We're building AI agents that, besides helping you create what you want, will help you and others in your meetings retrieve, organize, and think about your content and what's being discussed. We'll provide default instructions for these agents based on our years of experience building meeting software, but we'll also let you set your own instructions for these agents so that you can customize their behavior. There are a lot of beans I want to spill here, but I should probably save at least some surprises for later!

The fact that all this is "spatial" isn't the fundamental part of the value proposition. The reason it's spatial is because I think that's the best way to orchestrate and visualize interactions between people, AI agents, and content. We'll be making opinionated choices about the way the 3D environment is designed, with a specific eye towards facilitating the interactions with AI that add the most value for people who come together to collaborate online.

To sum up: the target market here is people and teams who have online meetings but who want to do so alongside AI-powered agents that have a deep contextual understanding of your organization and the topics of your conversations, and who can create, listen, challenge, understand, suggest, synthesize, and a lot more.

There's a longer term vision here too that we're shaping. I think a big part of the future of work will be teams comprised of humans and AI agents, working together towards shared goals. Some of this work will happen together synchronously, like in the meetings described above, but of course lots of work happens in between meetings.

Well, we're building AI agents that can do work inside your Frames whether you're in them or not. They will be able to work together in teams, autonomously or with humans in the loop. Let the implications of that sink in. They are vast.

I'll have a lot more to write about all of this stuff in the weeks and months ahead. I'm going to be actively "building out loud" as we cook this stuff up.

Hard Truths and Closing Words

If you just want to keep using Frame as it is and don't want to deal with any of this new stuff, you may still be able to. It depends on which features you rely on. You might not. The user interface will change, and some current features may get dramatically changed or removed. There will be new features.

While I hope that you'll stay on this journey and love what we're building, I owe it to you to be transparent: this will be a radical departure from the current Frame product. The truth here is that if we don't piss off at least some of our current users, we'll have been too cautious. We are going to be bold. It's too risky not to be.

There are so many enticing, juicy opportunities specifically at the intersection of AI and 3D. I'm overflowing with ideas. I don't think it's ever been a more exciting time to be building software.

I can't wait to show you our first steps towards our new vision.

Onwards and outwards,

Gabe