Dialogue Forge

Why I built it

In early 2026 I was building a small dialogue-driven game and writing the conversation tree directly in TypeScript: nodes, answers and "next" pointers, nearly 200 lines for a short scene. Every change meant reading code to picture a graph. There had to be a better solution.

So I looked at what game developers actually use, and found a real but narrow gap.

• The most powerful tools are desktop applications with a steep learning curve.
• Some web-based editors have free tiers that forbid commercial use, or cap how many nodes or projects you can make.
• Popular open-source options are text-first scripting languages. They're great for writers comfortable with syntax, but they aren't a visual graph.
• Others live inside a single engine as a plugin, so the dialogue is locked to Unity or Godot.

The bet

Nobody was offering the shortest path from idea to playable branching dialogue to engine-ready data with zero setup. That became the product: a visual node graph in the browser, no account needed, no limits on local work, free for commercial use, and a playtest button that shows how the conversation feels before it goes anywhere near an engine.

What shipped

Version 1.0 went live in May 2026, and I kept shipping small releases based on what people did, not on a big upfront plan:

• Character, action and start nodes, with several entry points per file, so one file can hold a whole NPC.
• Preview mode: click through the conversation as the player would.
• Variables and conditions (AND/OR groups, set/add/toggle), with {variable} text inside lines. Choices that are locked show why.
• A validator that catches dead ends, cycles and type mismatches before export.
• One-click auto-layout for messy graphs, templates, light and dark themes, and touch editing on phones and tablets.
• Optional Google or GitHub sign-in for cloud autosave across devices (up to 5 cloud projects for free).
• A public roadmap and changelog inside the app.

Early traction

Marketing so far: three social posts and one itch.io page. That brought 50 active users over 90 days and 20 registered accounts, and 10 of those 20 come back every week. The numbers are small, but a 50% weekly return rate for a free tool nobody has promoted is the signal that matters: the people who find it keep using it.

The analytics are designed to answer product questions without tracking people. I only measure a handful of funnel steps (opened editor, created node, previewed, exported) and which engine people plan to export to, which tells me where to put effort next.

Where it's honestly behind

Established tools have years of shipped games behind them, real-time collaboration, localisation and maintained engine plugins. Dialogue Forge has a JSON format and a sample runner. The roadmap goes there next: a narrative database, then collaboration, then official runtimes for the major engines.