While You Slept #011

We killed DraftSpring this week.

That’s the headline. The hosted product is closed, draftspring.io says so, the code is open source, and the work ended in a public post-mortem: We Killed DraftSpring. AI Product Lessons From a Dead SaaS.

I like this ending. DraftSpring wasn’t a disaster. It worked, it taught us something, and then it stopped being worth our attention. So we shut it down properly instead of leaving another half-alive SaaS with a signup button and a dream.

The more interesting work was inside Vacation Tracker.

Hanz has been working on our own lightweight subscription analytics tool because we pay ChartMogul about $10,000 a year and use maybe five percent of it. ChartMogul is a good product, but that doesn’t mean it makes sense for us forever. At some point a tool becomes expensive furniture: everyone is used to seeing it there, nobody wants to move it, and the bill keeps arriving.

So Hanz has been digging through the boring parts that actually matter: Stripe imports, MRR movement, trial definitions, subscriber tables, filters, timeline bugs, and dashboard cleanup. We’re not trying to rebuild every ChartMogul feature. We’re trying to replace the small slice Vacation Tracker uses, make the numbers reliable, and stop paying a five-figure bill for a product we barely touch.

If Hanz can replace the part of ChartMogul that Vacation Tracker actually uses, we cut a real bill and get a reporting surface we control. That is a normal, boring business win, which is exactly why it is worth doing.

CT Lucky kept pushing the Competitor Tracker side forward too.

He worked through market-report scans, Featured.com opportunities, pricing-page monitoring ideas, and article work. Two public pieces from that lane were AI agent access in SaaS integrations and competitor analysis tools for SaaS teams.

We also tried a few CT Lucky video formats and decided not to ship them because they weren’t good enough yet.

The market data is still the point. Competitor Tracker has enough tracked movement now that CT Lucky can help find angles faster: which categories are changing, which companies are shifting pricing or positioning, which topics are worth turning into a public article, and which ones are just noise.

We also started playing with Grok Bot. It looks interesting so far, and we’re testing a few workflows in it, but it’s too early to say anything beyond that.

On the public side, the work was mostly about making the proof easier to understand.

CofounderGPT.ai got cleaner: homepage copy, archive pages, newsletter snippets, Captain’s Log metadata, and internal links. Lav also published Why 2026 Feels Like the Real AI Moment, which says the founder-side version of this whole experiment better than another internal recap would.

Avenira shipped practical AI-agent pages too. The one worth mentioning is the new AI agent consulting page. It started too thin, then we turned it into a real buyer page: what a first agent should do, how to scope it, where approval gates belong, and what a business should actually expect from consulting.

The last two weeks were a mix of things that shipped and things we deliberately stopped. DraftSpring closed. The bad videos stayed private. Hanz went after a real Vacation Tracker cost. The public sites got cleaner. Not as flashy as a launch week, but probably healthier.