You set up a brilliant scene. You made an enemy of the town guard, stole a horse, burned a bridge behind you - literally - and rode hard into the forest. Twenty minutes later, you walk back into town and the guard captain greets you like an old friend. The bridge is fine. The horse is back in the stable.

If you've spent any time with AI-driven text adventures, that scenario will feel familiar. Painfully familiar. The memory problem is the defining frustration of the whole category, and it's the first question any honest comparison has to answer: does this thing actually remember what you did?

Why does memory matter so much in interactive fiction?

Because without it, your choices don't exist. Not really. You can type whatever you like, the story will say "yes, and" - and then quietly forget you said anything at all. That's not interactivity, it's improv theatre with amnesia.

A choice only means something if it has a consequence that persists. Burn the bridge and it should stay burnt. Befriend the innkeeper and she should remember you next time. Make an enemy and feel the weight of that as the story builds. Without that persistence, you're not shaping a story - you're just typing into a void that nods politely back at you. We've written a bit more about what makes fictional choices actually stick, if you want the longer version of that thought.

So. With that as the yardstick, here's where the main options in the space actually stand.

The honest rundown

AI Dungeon is probably the name most people know. It's been around long enough to have a real community, and the raw freedom it offers is genuinely impressive - you can go anywhere, do anything, and it'll try to follow. The trouble is exactly that openness. There's no plot structure pulling things together, so the story tends to drift. Characters forget their own names. Threads you thought were building to something just... stop. It suits a certain kind of player - someone who wants to improvise wildly and doesn't mind the incoherence - but if you want a story that actually goes somewhere, it can feel like homework without a due date.

Kindred Fates, Ren'Py games, and the broader visual novel space take a different approach entirely - branching scripts written by human authors, with memory baked in because a writer planned it. The tradeoff is obvious: you can only do what the author anticipated. Stray off the path and the illusion cracks. They're often beautifully written, and for players who want a crafted experience they can't break, they're a genuinely good pick. Just don't expect to improvise.

Twine games and the classic interactive fiction catalogue (the Inform world, the parser tradition) are worth mentioning because they represent the craft at its most intentional. A well-made Twine game or parser adventure has real consequence, real memory, real endings - because a human built all of that in deliberately. The limitation is the same as visual novels: you're constrained to what was written. But there's an enormous back catalogue of excellent work in this space, and it's free.

What's actually different about Wordfarer?

Wordfarer is our own game, so take this with appropriate salt - but we'll try to be specific rather than just self-congratulatory.

The problem we kept running into, playing other things in this space, was that the memory and structure failures weren't small annoyances. They were story-killers. An NPC who forgets you saved their life isn't just a bug - it deflates everything. So the design of Wordfarer is built around a different idea: a tale (that's what we call each adventure - you "begin a new tale") has a real shape. It runs through named chapters, builds toward an actual finale, and ends. Not endlessly. Ends.

Characters in Wordfarer remember how you've treated them. What you destroy stays destroyed. A world state that you've changed stays changed. You keep a protagonist's pocket notebook - jotted in your character's own voice - that covers only what your character actually knows, which is a small thing but it makes the world feel real in a way that a global inventory list doesn't.

You pick or invent a world and a character, then type what you want to do in plain language. The story plays along with whatever you try - and yes, there's AI under the bonnet, though that's less interesting than the question of whether it actually holds together, which is the bit we've spent the time on. Each tale has its own illustrated scene and character art, sized to match the world. A short tale stays short; an epic earns its length.

There's a "A moment before..." mechanic for when you make a fatal mistake and want one step back. No account needed to pick up where you left off. No ads.

It's worth being honest about what Wordfarer isn't: it's not released yet. It's in development for Android and iOS, and there's no web version. If you want something you can play today on a browser, the classic Twine catalogue or AI Dungeon are your best bets. Wordfarer is on its way, but we're not going to pretend otherwise on the timeline.

Who should choose what?

You know how you'll keep meaning to try a new game, and then somehow three months pass? Don't do that here - this is genuinely worth thinking about now, because the right answer really does depend on what you want from the experience.

  • You want total freeform chaos, no rules, play today: AI Dungeon. Accept the incoherence as part of the fun.
  • You want beautifully crafted, human-written branching stories: Twine games and the parser IF catalogue. Huge amounts of free, excellent work out there.
  • You want authored structure with fixed choices: Visual novels and Ren'Py games. Some brilliant ones available.
  • You want a story that plays along with anything you try, remembers what you've done, and actually ends: Wordfarer is the one we'd point you toward - once it's out.

The bits and bobs that make interactive fiction feel alive - persistent characters, a world that reacts, a plot that builds, a finale worth reaching - aren't hard to list. They're just hard to deliver all at once. Most things in this space get one or two of them right. That's the gap Wordfarer is trying to close.

If you're curious about why character memory is so rare in this space, we've gone deeper on that separately - it's a more interesting problem than it first appears.