I accidentally made myself obsolete

I built a feature for my users. Then I realized I built it for myself.

A few weeks ago, I realized something uncomfortable. I hadn’t opened the BlackTwist dashboard in days.

Not because the product was broken. Because I built something that made the dashboard unnecessary. For me, the founder.

Let me back up.

BlackTwist is a Threads scheduling and analytics tool. The dashboard is the product. You open it, write a post, pick a time, and schedule it. You check your analytics, see what worked, and plan the next batch. Standard stuff. I’ve been building and refining that dashboard for over a year.

Then MCP happened.

MCP stands for Model Context Protocol. It’s a way for AI assistants like Claude to connect to external tools. Think of it like giving Claude hands. Instead of just writing text, Claude can reach into other apps and do things. Schedule a post. Pull analytics. Set up automations.

I built an MCP Server for BlackTwist. The idea was simple. Let people manage their Threads account from inside Claude instead of switching tabs to the dashboard.

I tested it myself first. Obviously.

And I never went back to the dashboard.

The first day I used it, I was brainstorming a post in Claude. I had the angle, the hook, the text. Normally, I’d copy the text, open BlackTwist, paste it, pick a time slot, click schedule, switch back to Claude. Thirty seconds of friction that kills your momentum.

Instead, I typed: “Schedule this for tomorrow at 9 am on Threads.”

Claude checked my connected accounts. Checked my time slots. Checked for conflicts. Asked me to confirm. I said yes. Done. Never left the conversation.

That was week one.

By week two, I was planning my entire week from Claude. 21 posts. Three per day. Morning, midday, evening. Claude pulled my recent posts to match my voice, drafted everything in one document, and scheduled the full batch after I approved it. That workflow used to take me 3 hours on a Sunday. It now takes 12 minutes.

By week three, I was running weekly analytics reviews from Claude. It pulled my engagement data, built me an HTML dashboard as an artifact, and proposed content pillars for the next week based on what actually performed. One conversation. Five minutes. I got a chart showing my engagement dropped every Thursday through Sunday. I checked and realized I was posting personal stories early in the week and how to content late in the week. Personal stories were outperforming by 2x. I’d been running this account for a year and never noticed because I always looked at weekly totals.

Claude spotted it in one conversation because it was reading the data fresh.

Here’s what made me uncomfortable. I’m the founder. The dashboard is my product. And I stopped using it.

That’s a strange position to be in.

My first reaction was defensive. Maybe I’m just lazy. Maybe I’m too close to the AI hype. Maybe real users still need the dashboard.

But then I thought about it more honestly. I stopped using the dashboard because the new workflow is genuinely better. Not because the dashboard is bad. Because writing, scheduling, analyzing, and planning all happen in the same conversation. There’s no context switching. Claude already knows what I’m working on, what angle I’m exploring, what I rejected two drafts ago. The dashboard can’t know any of that.

And that realization changed how I thought about the product.

I had a decision to make. How do I tell people about this?

I could write a changelog post. “New feature: MCP Server. Connect Claude to BlackTwist.” That’s what most SaaS founders do. List the features, add some screenshots, publish, and move on.

But nobody reads changelogs. Especially not for a feature this abstract. “Model Context Protocol” means nothing to a Threads creator who just wants more followers.

So I did something different. I wrote about how I actually use it.

I ended up writing a 4-part series.

Post 1 covers connecting Claude to BlackTwist and scheduling your first post through a conversation. The 30-second setup.

Post 2 shows how to plan and schedule a full week. 21 posts, batched in one sitting, each in a specific format.

Post 3 is about automated follow-up replies. You schedule the post and the follow-up together. When the post hits a certain number of likes, BlackTwist fires the reply automatically. Whether you’re online or asleep.

Post 4 is the weekly analytics review. Claude pulls your data, builds a custom dashboard, and proposes content pillars based on what performed.

None of these posts mentions “MCP Server” in the title. They’re all framed as workflows. Problems solved, time saved, things you can actually do.

That was intentional. The feature is the MCP Server. The story is what it lets you do.

I’ve been building products for years, and I keep running into the same thing. The best marketing doesn’t feel like marketing. It feels like someone showing you how they solved a problem you also have.

I didn’t plan for the MCP Server to become a content series. I built it, used it, and the usage itself became the story. The documentation of my own workflow became the marketing.

That’s the loop I think every indie maker should look for.

If you’re building a product and struggling with marketing, stop thinking about marketing for a second. Ask yourself: do I use my own product? If yes, document that usage. Write it down. Every workflow, every shortcut, every moment where you thought “huh, that was easier than expected.”

That document is your marketing.

If you don’t use your own product, that’s a different problem. Fix that first.

If you’re curious about the actual workflows, here’s the full series:

1. Schedule Threads posts from Claude without touching a dashboard

2. Plan and schedule a full week of content from one conversation

3. Set up automated follow-up replies on your posts

4. Run a weekly analytics review with a custom dashboard

Each one is a standalone walkthrough. You can read them in order or jump to whichever problem you want to solve first.

And if you just want to try the MCP Server, blacktwist.app/mcp

See you next week.

Luca