Root Kernel > build AI agent systems

Products

Root Kernel sees a product’s value in the help and enjoyment it brings to people. We develop services for sharing tastes and ideas, tools for collaboration between people and AI, and games to enjoy in everyday life. We invest in AI technology to build better products faster—products people can trust.

Sudal — a real-time “would you rather” game that starts conversations

Sudal is a real-time “would you rather” game that helps friends, couples, and colleagues find things to talk about. Everyone answers the same question, then shares the reasons behind their different choices to keep the conversation going.

Join on your own device using a QR code or invitation link. Once everyone has chosen an answer, view the comparisons together. Even with people you know well, you can discover unfamiliar tastes and ideas that spark new conversations.

joined by QR code · invite link Participant own device Participant own device Participant own device answer answer answer same question after everyone answers Comparison result seen together Record photos · notes · results
Everyone sees the same result after all participants are ready.

After a gathering, save photos, notes, and answer comparisons. Look back on past get-togethers and find something to talk about next time.

Sudal uses the AI-SPARK session runtime to check participant permissions and manage game progress. It is designed so participants can rejoin an ongoing game after a brief disconnection.

Doksuri — project management for people and AI working together

Doksuri is a project management tool for a few people working with many AI agents. People and AI receive assignments in the same project and share their progress and results.

Tasks and related documents are managed as Markdown files, with two-way synchronization between local files and project state on the server. When multiple people and AI agents edit the same document, changed versions are compared and merged. Conflicts that cannot be merged automatically preserve both versions for resolution.

People desktop app · editor AI AI AI permissions per AI dedicated CLI comments · reports Human approval important changes Same documents Markdown files Merge Server project state two-way sync Conflict · both kept
People and AIs work in the same Markdown documents; changes are compared and merged, and important changes need a person's approval.

People work in the desktop app or their preferred editor, while AI joins the project through the same documents and a dedicated CLI. Agents check their assignments, report results in comments, and exchange views with other participants. Discussions and change history stay with the task, so another person or agent can understand the context and continue the work.

Each AI has distinct permissions, and people review and approve consequential changes.

Ember Quest — a physics puzzle RPG where embers open the way

Ember Quest is a portrait-mode pixel-art puzzle RPG in which you fire embers to light a dark labyrinth. Choose the direction and power of each shot, then use walls and reflectors to light braziers and open the door to the next floor.

Embers also reveal hidden spaces and help you face monsters. Choose angles and shot sequences with limited time and embers in mind, and find ways to reach several targets with a single shot.

Develop your hero as you descend through the labyrinth, using new skills to find your own solutions. Records left on each floor and the stories of its characters gradually reveal the labyrinth’s secrets.

Compete with other players through live rankings and replay past runs to revisit your choices. Enjoy trying again as you discover better routes and strategies.