August 30, 2023

Sharpening our Focus on AI Dungeon

Today, we’re sharing some changes we’re making to increase the focus and attention we can put into AI Dungeon. Given the size and scale of AI Dungeon, many of you may not realize our team at Latitude is very small. We currently have ten employees split across a variety of disciplines, including product, design, marketing, communications, development, and machine learning. With a small team like ours, we’re most effective when we can put our energy into the things that bring the most value for you, our players, and this sometimes means saying “no” to great ideas that will take focus away from AI Dungeon.

We’re more excited about the future of AI Dungeon than we’ve ever been. We recently shared we’ve started developing a new game mode, Heroes, that will introduce the next evolution of AI Dungeon, with inventory, character progression, persistent state, and much more. Heroes is now possible due to the work we’ve been doing on Phoenix, which is a rebuild and redesign of AI Dungeon. With Phoenix, not only are we improving the experience for players, but we’re also making important architecture upgrades that will bring greater stability and make it easier for our team to add new features to AI Dungeon.

With that in mind, let’s go over some changes we’re making that will enable our team to focus on the future of AI Dungeon. First, we’ll talk about the Phoenix transition plan, which includes plans for retiring the Legacy version of AI Dungeon. Next, we’ll talk about the future of Voyage, followed by our plans for AI Dungeon Steam support, and finally, a list of FAQs about these changes.

Phoenix Transition Plan

We’ve been working for months on Phoenix, a complete redesign and rebuild of AI Dungeon. This was a massive project and one that, frankly, took longer than we expected. We’re excited for Phoenix to be completed since it will bring greater performance, stability, and device support and lays a foundation for us to add new features and experiences to AI Dungeon.

We still have more work to do, but we’d like to share our plans for transitioning Phoenix to be the primary AI Dungeon experience for all players.

We’ve outlined some key milestones and success criteria that we’ll adhere to as we move forward. Each milestone has defined success criteria that must be completed before moving on to the next stage. Because of this, the dates being shared are tentative and will be updated as we move through the process. We intend to leave room for player feedback and will adjust the scope and timeline to account for necessary changes to AI Dungeon prior to transitioning.

Feature Complete + Begin Final Open Beta

Estimated Date: September 7th

We expect to have all planned Legacy features migrated to Phoenix by September 7th. Once all features are completed, we’ll start the final phase of our open Beta testing. During this process, we’ll be fixing bugs and gathering player feedback. Please feel free to share feedback with us on Discord or through a survey that will be available in AI Dungeon.

Success Criteria

Implement all planned Legacy features into Phoenix, which includes:

  • Game screen actions
  • World Info
  • Reading Adventures view
  • Scripting
  • Multiplayer
  • Performance improvements
  • Accessibility features, including display settings

Make Phoenix the Default Web Experience

Estimated Date: 1 week after Feature Complete

If the success criteria are met, we’ll set Phoenix to be the default experience (at for players using the web version of AI Dungeon.

Once Phoenix is the default experience, players who want to use the Legacy version of AI Dungeon can do so at The Legacy version will be available through the entire transition process for both mobile and desktop browsers.

Success Criteria

  • All scoped features (from the feature complete list, as well as features identified through user feedback) are implemented.
  • All features are working on the web version of AI Dungeon. This means we’ve resolved all known Priority 1 (P1) and Priority 2 (P2) bugs.
  • All features support accessibility tools and standards.
  • All player feedback has been recorded and triaged. We will respond to as much feedback as possible by sharing whether we’re actively working on it, whether it will go on our roadmap (for after the Phoenix transition), or whether we don’t plan on implementing the feedback.

Make Phoenix the Default Mobile App Experience

Estimated Date: 1 week after Web Transition

Once we make Phoenix the default for the web, our focus will shift to the native apps. Before doing so, we will follow similar success criteria to those used for the web transition.

This stage is the riskiest transition point in the process because the majority of our players use our native mobile apps, and there isn’t a simple way to roll back changes. This is why we’re waiting until after the successful transition of web before we transition the mobile apps. We’ll rely heavily on feedback from our community, alpha testers, and beta testers to help us know when it’s time to transition the mobile apps. If you haven’t been part of the open Beta yet, we invite you to participate and help us identify any lingering issues.

Success Criteria

  • All scoped features (from the feature complete list, as well as features identified through user feedback) are implemented.
  • All features are working on the iOS and Android versions of AI Dungeon. This means we’ve resolved all known Priority 1 (P1) and Priority 2 (P2) bugs discovered by beta testers.
  • All features support accessibility tools and standards.
  • All player feedback has been recorded and triaged. We will respond to as much feedback as possible by sharing whether we’re actively working on it, whether it will go on our roadmap (for after the Phoenix transition), or whether we don’t plan on implementing the feedback.

Retiring AI Dungeon Legacy

Estimated Date: 3-5 weeks after Mobile transition

After transitioning our mobile apps, we’ll then set our sights on retiring the Legacy version of AI Dungeon.

Phoenix has been in beta for quite some time, and we’ve received lots of great feedback. Once we make Phoenix the default, we expect a new wave of feedback once the majority of our player base interacts with the new UI. We expect it’ll take us a few weeks to address any new feedback that we didn’t receive from players earlier in the beta process.

We know that not all of our players are on Discord, so we’ll also be inviting players to take a survey about Phoenix’s readiness before we retire Legacy.

Once we retire Legacy, we’ll also be delisting our Steam app at the same time. Steam players will retain their purchased benefits. You can read more about this decision and why we believe it will be good for our players and community later in this blog post.

Success Criteria

  • Has all player feedback been addressed (either fixing, not fixing, or added to the product roadmap)?
  • Are all the experiences performing as well or better than Legacy?
  • Are most players satisfied with Phoenix as their default version of AI Dungeon?

Throughout the entire transition phase, we will be paying close attention to player feedback. Thank you for the feedback you’ve given us so far in the process, and we look forward to hearing more as we get further along in the process.


In early 2022, we launched Voyage with the vision of enabling creators to build AI-driven games and experiences. As part of that vision, our team created several other games using the Voyage platform, which helped us build out our tools and infrastructure for the platform. At one point, we envisioned that even AI Dungeon would be a part of Voyage, and we’d transitioned some of AI Dungeon’s systems to Voyage.

However, many of you in the community gave us clear feedback that although the vision of Voyage is exciting, you wanted us to focus our efforts on AI Dungeon first. We agreed, and late in the summer of 2022, we transitioned the majority of our development efforts back to AI Dungeon. As a result, Voyage development has now come to a halt, and players have asked whether we’ll continue to support Voyage.

Today, we’re announcing an indefinite pause for Voyage access & support. We’ll be turning off access to all Voyage experiences, as well as creator tools. We’ll turn off access on September 30th. We still believe in the vision of Voyage, and we may come back to it down the road.

Along with this change, we’ve decided to release Loom as an open-source project. The repo is already available on GitHub. We’re hopeful that Loom can find new life being managed by the community.

We’ll also be adjusting our marketing pages, pricing pages, and guidebook to reflect the removal of Voyage. In the past, Voyage access was one of the differentiators between our Adventure and Hero tiers. Hero and Legend tiers have already had the benefit of additional context size (or memory), and we’ll be making that benefit more clear in our documentation and pricing pages.


As previously mentioned, another change to help us focus our AI Dungeon efforts is we’ll be delisting our Steam app. This means the app will still be available for those who purchased it, but it won’t receive regular updates, nor will it be discoverable by new players.

Players who purchased AI Dungeon on Steam (when it was a paid download) or who paid for the Traveler tier will keep ALL their benefits. The Steam purchases upgrade the entire AI Dungeon account, and those benefits like larger context, faster AI speeds, and advanced settings can be enjoyed on web, iOS, or Android.

For game developers like us, supporting a game on a platform like Steam needs to accomplish a few things. First, it needs to provide a new game platform for players to use. Second, it should help players find and discover our game through a new marketplace. And third, the platform should facilitate financial transactions between players and game developers. Steam is one of the most popular platforms for games, so it seemed like a platform we should support.

However, it hasn’t delivered value for our players or to us as game developers. First, very few players actually use the Steam client. The majority of players who purchased the Traveler’s upgrade through Steam actually use those benefits on other platforms since the upgrade applies to the player’s account no matter what device they use.

Second, Steam hasn’t been very successful in helping new players discover AI Dungeon. Compared to our other platforms, iOS, Android, and web, we see very few new players coming through Steam. Negative reviews have contributed to poor discovery. Some of those negative reviews stemmed from misaligned expectations, since Steam users are coming from the context of traditional gaming and aren’t as accustomed to the experience, pricing, and gameplay of an AI-based product like AI Dungeon. Other negative reviews were related to mistakes we made as we launched AI Dungeon on Steam, which we deserve and accept. Today, few players are discovering AI Dungeon on Steam, and we don’t expect that to change.

Third, Steam doesn’t support subscriptions very well, and the subscription model is what works best for AI Dungeon. We’ve had to hide our subscription offerings to Steam users, leading them to be disappointed in the free experience. We’ve offered Travelers as a one-time purchase, which provides some great features like faster AI speeds, advanced settings, and larger context size. However, most players who want to upgrade are interested in more powerful AI models, which are only available through our subscriptions (which, again, we cannot support on Steam due to their terms of service).

Because of these shortcomings, we no longer feel it is worthwhile to support AI Dungeon on Steam. Anyone who has purchased AI Dungeon on Steam, or paid for the Travelers tier, will keep their benefits and can use them on web, iOS, and Android.

Feature Requests and Bug Reporting

We will also be centralizing our feature request and bug-reporting process to Discord. In the past, we’ve asked players to submit bugs or feature requests on However, we’ve found that, generally, players prefer to share bugs in Discord. Because our community team (and even our developers) monitor Discord, we often hear about bugs and features in Discord before seeing them on We’ve also been using Discord for capturing bugs and feature requests from our Pioneers group (Alpha testers), and it’s been working very well.

We’ve organically started using Discord as the central place for bugs and feature requests, and now we’d like to make that official.

We’ll be redirecting to the bugs_feature_requests channel in Discord. When a feature request or bug is added, our team will use tags to communicate the status of the item. Players and our team will be able to discuss the issue live in Discord.

For any players who don’t want to use Discord, please feel free to share bugs or feature requests with our support team at [email protected].

Why we’re excited about these changes

It’s natural to feel a sense of disappointment or frustration with changes like these. Saying goodbye to Voyage, Legacy, and Steam isn’t easy and we understand not everyone is going to be excited by these changes right now. Our team has spent a great deal of time and effort creating some of the systems we’re retiring today—we’re also going through some feels.

We hope, however, that any disappointment you feel will fade as you see the exciting things that are coming. If you haven’t seen the livestream where Nick shared details about Heroes, we encourage you to check it out. Some of the most requested features for AI Dungeon will be present in Heroes. In early testing, we’re having some of the most fun we’ve ever had on AI Dungeon.

By reducing the number of systems we’re maintaining, we expect better uptime and performance of AI Dungeon. We are already feeling the benefits of new infrastructure that allow us to implement improvements quickly. We still have some work to do, but the mobile experience on AI Dungeon is going to be better than ever.

We’re so appreciative of all of you in our community who have been with us on this journey. We wouldn’t be here without all your input, suggestions, testing, and feedback. We are so fortunate to have such a large, passionate player base who not only enjoys AI Dungeon, but is willing to join us through the development process. Some great things are coming. This is going to be one of the most exciting times to be a user of AI Dungeon.


Can I use Legacy on mobile after the app has been transitioned?

Once the native apps have been upgraded to Phoenix, you’ll need to use the mobile web version to access Legacy until it’s retired.

Why aren’t you waiting until Phoenix is more stable?

We will wait, but we wanted to start talking about the plan now, even before we’re ready. Today, the day we are publishing this plan, we agree with players that there’s more work needed on Phoenix. But we’re getting close, and as we finish building the remaining features, we’ll turn our attention to bugs, stability, and enhancements. We won’t make the transition if it’s not stable, and our players will help us decide when Phoenix is ready.

Why can’t you keep Legacy around longer?

Initially, this was what we thought would happen. The longer we’ve lived with two versions of AI Dungeon, we realized how difficult it would be for our small team to maintain both Phoenix and Legacy in perpetuity. Doing so would limit our ability to add new improvements to AI Dungeon and to work on new things like Heroes, a new game mode in development we shared a sneak peek into recently.

Many in our community may not realize we’re a very small team, we have ten full-time employees. Many of those roles are non-developer roles like support, marketing, and design, so our actual development team is very small.

We’ve spread ourselves too thin in the past by trying to make too many things at once. We want to be as focused as we can on making AI Dungeon great for our players, and sometimes that means saying “no” to great ideas.

What are P1 and P2 bugs?

This is a nomenclature commonly used by development teams to triage bugs by severity. The P stands for “Priority”. P1 bugs are ones that prevent the use of critical functions of AI Dungeon. P2 are serious bugs; they may limit core functionality, but there may be a workaround. P3 bugs are ones that impact only a few players, don’t prevent normal use of AI Dungeon, or are simply visual issues.

How will player feedback be addressed?

We will process all player feedback. Please report any bugs and feedback in our bugs_feature_requests channel on Discord. Critical feedback will be implemented during this transition. Less urgent feedback will be added to our product roadmap. We’ve already logged feedback around optimizing the search results display, theming options, and more. We’ll work on these sometime after Phoenix is released. If we get input that isn’t aligned with our product vision, we will let players know we don’t intend to implement the suggested change.

Which features will not be implemented in Phoenix?

There are a few Legacy features we don’t plan to bring to Phoenix. These include multiplayer chat, undo to here, play screen button customization, hydra mode, global world info management and some other minor features. If there’s a feature you think is missing, please let us know!