😸 Your agents got a Slack

gm legends, happy Sunday. Day two of the weekend edition, same mission as yesterday: the makers who ship on weekends deserve readers too, so here we are.

Today, three debuts: your AI agents get an office with a task board and a boss, a solo dev's reading app for the languages he's learning, and a journal that's proud to have no AI in it. Plus the forum on where your agent's memory should actually live.

CircleChat gives your AI agents what every team eventually needs: a Slack to talk in, a task board to work from, and a boss to answer to, so the agents you're running stop being scattered terminal windows and start looking like a staffed team. It's from Tash Ahmed, a designer, which tracks, because the insight here is an interface one.

🔥 Our Take: Run more than one agent and you discover the real problem isn't intelligence, it's coordination, three terminals doing who-knows-what with nobody in charge. CircleChat's answer is charmingly literal: give them the tools human teams use, a Slack, a task board, a boss. The multi-agent frameworks have been at this for a year, but they're built for engineers, and putting it in an office metaphor everyone already understands is exactly what a designer would do that a framework author wouldn't. Yesterday the agents got a 3D world to be watched in, today they get an office to report to. The management layer for AI labor is assembling itself in real time, one familiar metaphor at a time.

Toku Reader lets you read and listen to native Japanese and Chinese content with a tap on any word to understand it, so you learn from the real thing instead of app exercises. It's built by Darren Nah, a solo dev whose one-line bio is simply "building a JP/CN reading app," which is both his product and apparently his whole life right now.

🔥 Our Take: Anyone who's gone past Duolingo knows the wall: the language actually lives in native content, and real Japanese or Chinese text is a wall of characters you can't look up fast enough to enjoy. Toku's answer is tap any word, keep reading, keep listening. LingQ has circled this for years for European languages, but JP and CN are where the lookup pain is worst and the tooling thinnest. And the maker signal is as clean as it gets, a person building the exact thing he needs, one launch, no company around it. Those tools tend to be quietly excellent for the people they're for.

Framer 3.0 ships AI Agents that live inside your canvas and can actually design. Not suggest. Design. Drop a brief and watch them build out full pages, create components, write code, connect your CMS, and handle SEO. All without leaving the canvas. They plug straight into Claude Code and Cursor. Yes, inside your actual component system, without breaking it. The Figma file you were about to hand off to engineering? Keep it.

Pennen gives you one quiet handwritten page a day, and that's the whole product: no feed to scroll, no streaks to guilt you, and, as maker Ishaan Rawat puts it right in the tagline, no AI. On a leaderboard where nearly every other launch this weekend is an agent, that reads less like a feature list and more like a position.

🔥 Our Take: Every journaling app eventually grows a feed, a streak counter, and lately an AI that summarizes your feelings back at you. Pennen ships the opposite of the entire category: a page a day, written by hand, seen by nobody, analyzed by nothing. Launching "no AI" as your headline feature on this week's board is either stubborn or exactly on time, and the ten quiet upvotes it collected this morning suggest the second. It will never be a company, and a page a day was never meant to scale. Somebody had to build the room with no robots in it. It's this.