AI that understands your code, not just your tickets

7 AI actions built for developer workflows

Kojit's AI isn't a generic chatbot. It's 7 specific, workflow-integrated actions designed for software projects:

1. Project Overview Generation — Analyze your codebase, commits, and canvas to produce a comprehensive project summary including tech stack, architecture, key features, and AI suggestions.

2. Commit-to-Canvas Analysis — Automatically transform your commit history into categorized roadmap nodes (features, fixes, milestones) with one click.

3. Smart Canvas Layout — AI analyzes node descriptions and relationships to suggest optimal spatial arrangements, grouping related items semantically.

4. Release Changelog Synthesis — Generate professional changelogs from your merged PRs and releases.

5. Node Deep Analysis — Get detailed insights on any canvas node: dependencies, related commits, scope assessment, and recommendations.

6. Docs Writing Assistant — Continue writing, improve text, summarize, or generate project documentation from your commit history.

7. Prompt Enhancement — Generate enriched prompts from node or section context, ready to paste into any AI coding assistant.

Commit analysis: from code to roadmap automatically

The commit-to-canvas AI action is Kojit's most powerful feature. It analyzes your recent commit history and automatically:

• Groups related commits into logical features and workstreams • Creates categorized nodes (feature, bug fix, refactor, milestone) on your canvas • Generates descriptive titles based on commit message analysis • Preserves commit links so you can always trace back to the code

This means your roadmap builds itself from your actual development activity. No manual entry, no guesswork — just a visual representation of what your team has actually built and shipped.

Smart layout: let AI organize your canvas

As your canvas grows, keeping it organized becomes challenging. Kojit's AI layout action analyzes the semantic content of your nodes — descriptions, linked commits, types, and dependencies — to suggest intelligent groupings.

The AI might group all authentication-related nodes together, separate frontend from backend work, or cluster items by sprint. The layout is applied with one click, and you can undo it instantly if you prefer a different arrangement.

Pro users get 5 AI layouts per month; Team users get 20. Each layout considers the current state of your entire canvas for the most relevant organization.

AI docs assistant: write less, document more

Kojit's docs editor includes an AI writing assistant that understands your project context. Unlike standalone AI writing tools, it can:

• Generate documentation pages from your project's commits, releases, and canvas data • Continue writing from where you left off, matching your style and context • Improve existing text — fix grammar, shorten, expand, or summarize • Translate between French and English • Insert project-specific data: recent activity, feature lists, technical details

The assistant is available via slash commands in the docs editor. Type /ai and describe what you need. The AI generates content using your actual project data as context.

Privacy and control

Your code stays yours. Kojit's AI processes commit messages, node descriptions, and canvas structure — not your source code files. All AI processing happens through encrypted API calls to OpenAI, and no data is stored or used for training.

Every AI action is human-in-the-loop: you review, edit, and approve all AI-generated content before it's applied to your project. Nothing changes without your explicit confirmation.