Todo.is

Accepts freeform natural language input through a chat interface and parses it into structured task objects with title, description, due date, priority, and assignee fields. Uses NLP to extract temporal references (e.g., 'next Friday', 'in 2 weeks'), priority signals ('urgent', 'low-key'), and implicit task structure from conversational phrasing. The system likely tokenizes input, applies intent classification, and entity extraction to populate task metadata without requiring manual form filling.

Analyzes task attributes (due date, description keywords, project context, team velocity) and user behavior patterns to assign or suggest priority levels and urgency scores. Likely uses a scoring function that weights factors like temporal proximity ('due tomorrow' = high urgency), keyword signals ('critical', 'blocker'), and historical task completion patterns. The system may employ collaborative filtering to infer priority from similar tasks completed by other team members.

Enables multiple team members to view and edit the same task simultaneously with live updates, cursor presence indicators, and conflict-free concurrent edits. Likely uses operational transformation (OT) or conflict-free replicated data types (CRDTs) to merge concurrent edits without requiring explicit locking. The system broadcasts presence state (who is viewing/editing which task) and updates task state across all connected clients in near-real-time via WebSocket or similar persistent connection.

Maintains a multi-turn chat context where users can ask the AI to clarify, expand, or break down tasks into subtasks through natural language. The system retains conversation history and task context, allowing users to say 'split this into smaller steps' or 'what are the acceptance criteria?' and receive AI-generated suggestions. This likely uses a retrieval-augmented generation (RAG) pattern where the current task and conversation history are injected into the LLM prompt to generate contextually relevant suggestions.

Analyzes task attributes (skills required, project context, team member workload, historical assignments) and suggests optimal assignees or automatically routes tasks to team members. The system likely maintains a skill matrix or historical assignment log, uses workload balancing heuristics to avoid overloading individuals, and may apply collaborative filtering to match tasks to team members with similar past assignments. Suggestions are presented to the user before assignment to maintain human oversight.

Provides a free tier with core task management functionality (create, view, edit tasks; basic collaboration) and gates advanced AI features (prioritization, assignment suggestions, decomposition) behind a paid subscription. The system likely tracks feature usage and API calls (LLM inference, prioritization scoring) and enforces rate limits or feature availability based on subscription tier. Free tier users can access the product without credit card, reducing friction for individual adoption.

Maintains a chronological log of all changes to tasks (edits, assignments, status changes, comments) with timestamps and attribution to specific users. The system displays this activity feed in the task detail view, allowing team members to understand the evolution of a task and who made what changes. This likely uses an event-sourcing pattern where each change is recorded as an immutable event, enabling both real-time updates and historical queries.

Allows users to search and filter tasks using conversational queries (e.g., 'show me all high-priority tasks due this week assigned to Sarah') rather than requiring structured filter syntax. The system parses natural language queries into structured filter expressions (priority=high, due_date<=next_week, assignee=Sarah) using NLP entity extraction and intent classification. Results are returned as a filtered task list with optional sorting and grouping.