July 2026
More ways to create.
July 2026 performance snapshot
I have now fully defined the set of metrics I will use to evaluate the success of this product.
First and foremost, I want the product to support the Learning Loop we put our students into: a student creates a project, shares it to get feedback, and iterates on it based on that feedback. To track this loop, I use the following metrics (here and below, compared to June 2026):
Publishes per active student: 0.7 → 1.4
Reactions per published project: 0.3 → 0.7
Publishes after modification share: 0.30% → 12.90%
The second most important group of metrics relates to gamifying this loop. The key idea behind the gamification is that since publishing a project is at the center of the learning loop, profile customization rewards become valuable to students. They help them make themselves and their projects stand out in the feed. In addition, teenagers naturally want to both express their individuality and feel like part of a group through external attributes, such as clothes, hairstyles, merch, etc. I measure the effectiveness of gamification by tracking the use of customization features and the share of students who receive rewards:
Share of users who use customization features: 47.4% → 67.5%
Share of users who open the reward box: 9.6% → 17.5%
Of course, it is also important to track audience size:
MAU: 1,745 → 2,282
As these metrics show, the July releases definitely made the product better.
July releases
More ways to create
I have significantly improved the studio UI and added several new tools requested by our curriculum team.
Here are the key updates:
7 creative studios available for students: JS and Python web projects, Python terminal, pixel art and animation, Minecraft skin editor, music, and Scratch.
Major UI improvements in Studios: updated toolbars, new keyboard shortcuts, full-screen mode, automatic project thumbnails, and more.
New project type — Story: block-based project layout, a media gallery, and PDF export.
AI that guides
I designed a two-agent AI system with different purposes and a shared memory that turns on only with the student's explicit consent.
Two agents — two roles
Robby (Tutor):
Where: inside Studios, Challenges, and project feedback.
Role: teaches and guides, but strictly does not write code for the student.
Context: sees live project files, tech stack, and the task list.
Aiko (Helper):
Where: on a dedicated page (/ai-helper).
Role: a general assistant that completes specific tasks and provides ready-made solutions upon request.
Context: conversation history, pinned files, and ideas.
Shared memory and privacy
The agents share a common memory about the user and save facts there automatically. This memory is turned off by default and activates only after the student clearly confirms it. In addition, students can see everything the agents have remembered, edit the information, or delete it completely.
Platform & lessons
Together with the school's IT team, I improved the student journey (CJM) to make using the product as smooth and seamless as possible. Students now have access to:
One-click seamless login: the portfolio service works separately, but students no longer need to log in again — login happens automatically if they are already signed in to the school services.
Embedding tools directly into lessons: individual product components can now be opened directly inside the Kodland learning platform. During a lesson, users can launch a specific studio to work on a project or open an AI assistant (like Aiko) for quick help.