Why I added AI to SteamLaunch (and what it actually does)

I built SteamLaunch to organize my Steam launch. It has a roadmap, deadlines, content calendar, the whole thing. It works well as a web app. I use it every day.

But I also use Claude every day. For writing devlogs, brainstorming content ideas, figuring out timelines. And I kept running into the same friction: Claude doesn't know what's on my roadmap. It doesn't know my launch date. It doesn't know what content I've already scheduled. Every time I wanted help with something launch-related, I had to re-explain my entire situation.

After a few weeks of this I decided to fix it.

The problem with context switching

Here's what my workflow looked like before: open SteamLaunch to check what deadline is coming up. Copy that into my AI chat. Ask it to help me figure out what to focus on. Then go back to the browser. Open the content calendar. Create a new entry manually. Back and forth, multiple times a week.

It's not that any single step was hard. It's that I was doing this dance multiple times a week, and the AI never remembered anything from last time.

What MCP is

MCP stands for Model Context Protocol. It lets AI assistants connect to external tools and read or modify data. In practice: your AI gets a set of functions it can call. "Read this person's deadlines." "Create a content entry for June 15." "What's my wishlist velocity?"

Claude Desktop, Cursor, and Claude Code all support it. You add a configuration block with a URL and a token, and the AI can talk to your app directly.

What I built

24 tools spread across three areas.

Content. The AI can read your content plan, import a new plan (I use this with Claude to generate a whole editorial calendar in one conversation), create individual content items, and get SteamLaunch's suggestions for what to post next.

Checklist. List tasks by phase, mark them complete, create custom tasks. I use this when I'm in Claude and remember I finished something. "Mark the capsule art task as done." Done. No need to open the browser.

Project data. The AI can read your full project status: launch date, wishlist count, velocity, upcoming deadlines, goal progress. When I ask "what should I focus on this week?" it actually knows my situation. It sees that Next Fest registration is in 18 days and my demo isn't submitted yet. No more re-explaining everything from scratch.

How I actually use it

The biggest use case is content scheduling. I use the prompt template (SteamLaunch generates one with my dates pre-filled) to have my AI lay out a publishing schedule: what to post, on which dates, on which channels. Then I import that schedule into SteamLaunch and it tracks everything. Five minutes instead of building it manually in the calendar.

The second use case is the weekly check-in. "How am I doing?" Claude pulls my dashboard summary and gives me a straight answer: your velocity is on track, but you have a deadline in 12 days that you haven't started on. That's useful context I'd otherwise have to go look up myself.

The third is small tasks throughout the day. Log a wishlist count. Schedule a post. Mark a task done. Things that take 30 seconds in the browser but 5 seconds in a conversation.

Setting it up

It's one token and one config block. Go to Settings, generate an API token, copy the config for your AI client, paste it. That's it. No OAuth flows, no API keys to manage, no complicated setup.

The token is yours. You can revoke it anytime. The AI only accesses your data, never anyone else's.

Why it's optional

Most devs don't use AI assistants for their workflow, and SteamLaunch works perfectly without it. The MCP part is there for people who already spend time in Claude or Cursor and are tired of switching tabs.

I added it because I needed it. If you don't, ignore it. The roadmap, deadlines, content calendar, and guides all work the same with or without AI connected.