Ohai.ai

Converts unstructured text messages into actionable household tasks by parsing natural language intent, extracting entities (items, assignees, deadlines), and creating structured task records without requiring explicit formatting. Uses NLP to disambiguate context (e.g., 'we're out of milk' → add milk to shopping list) and infer task type from conversational phrasing rather than requiring users to select categories or fill forms.

Maintains a persistent, queryable knowledge base of household state (who's responsible for what, current inventory, recurring patterns, family preferences) built from conversation history and task completion data. Uses retrieval-augmented generation to surface relevant context when processing new requests, enabling the AI to make informed decisions without re-asking questions (e.g., remembering that Sarah always handles grocery shopping).

Tracks household expenses mentioned in conversation (e.g., 'spent $50 on groceries') and maintains a budget ledger with optional categorization and spending alerts. Implements expense recognition from natural language mentions and can provide spending summaries or budget status updates when queried, without requiring users to manually log expenses in a separate app.

Orchestrates task distribution across household members by parsing natural language requests, inferring appropriate assignees based on historical patterns and stated preferences, and creating accountability through shared visibility. Implements a task routing system that can assign work based on availability signals, past responsibility, or explicit delegation without requiring manual assignment UI interactions.

Aggregates shopping items mentioned across multiple text conversations into a unified, deduplicated shopping list by recognizing item mentions in natural language (e.g., 'we're out of milk', 'need more pasta'), merging duplicates, and organizing by store section or priority. Implements fuzzy matching to detect when 'milk' and 'whole milk' refer to the same item, and allows users to update the list via continued conversation rather than explicit list editing.

Detects recurring household tasks from conversation patterns (e.g., 'we always need milk on Sundays') and automatically schedules reminders or task creation on inferred cadences. Uses temporal reasoning to understand frequency mentions ('weekly', 'every other Thursday', 'monthly') and creates automated task generation without requiring users to set up recurring tasks explicitly.

Tracks task completion status across household members and surfaces accountability metrics (who completed tasks, who's behind, completion rates) through conversational queries. Implements a completion state machine (assigned → in-progress → completed) and allows users to update status via text (e.g., 'done with laundry') rather than clicking checkboxes, with optional notifications to other household members when tasks are completed.

Integrates with multiple messaging platforms (SMS, WhatsApp, iMessage, Slack, etc.) to provide a unified interface where household members can interact with the AI through their preferred communication channel. Routes all household coordination requests to a single backend system regardless of input channel, and broadcasts responses back through the same channel or to all household members depending on message type.

Learns household-specific preferences, dietary restrictions, communication styles, and task preferences from conversation history and explicit user input, then personalizes task suggestions, reminders, and responses accordingly. Implements user profiles that capture individual preferences (e.g., Sarah prefers morning reminders, John is vegetarian) and uses these to tailor system behavior without requiring explicit configuration.

Allows household members to query the system conversationally to understand current household state (e.g., 'what's on the shopping list?', 'who's responsible for laundry?', 'what do I need to do today?'). Implements a natural language query interface that translates conversational questions into structured queries against the household task and state database, returning human-readable summaries rather than raw data.

Integrates with household calendars (Google Calendar, Apple Calendar, Outlook) to understand availability and schedule-dependent tasks, and allows users to coordinate events through conversational requests (e.g., 'schedule a family dinner for when everyone's free'). Uses calendar data to infer availability windows and suggest optimal times for household activities without requiring manual scheduling.