TwinMind

By TwinMind

Get New App Alerts

Open in ChatGPT

Open in Claude

Overview

TwinMind captures everything you say and hear in your meetings, lectures, and in-person conversations. It automatically transcribes, summarizes, proactively offers suggestions and tasks, and continuously builds context for your AI Second Brain. TwinMind is a productivity tool leveraging AI to function as your second brain. It captures everything heard, seen, and said—growing with you the more you interact and engage with TwinMind to build a comprehensive context of your digital life. This captured context enables several key features: Personalized and Proactive Assistance: TwinMind provides tailored suggestions and ideas based on your captured memories, anticipating needs and offering relevant insights. Streamlined Productivity: It enhances your notetaking by having your AI assistant automate tasks such as generating summaries and action items from meetings and conversations. You can search their entire memory bank, ask questions, generate emails and study guides, investor memos, project plans, and more. Privacy Focus: TwinMind prioritizes user privacy through on-device transcription, options to save transcripts locally or on the cloud. We're committed to zero third-party data sharing. Audio is processed on-the-fly and never stored. Key Features and Functionality: Capture and Context Building: Captures your activity (audio, visual, text) on mobile and desktop if partnered with the TwinMind chrome extension. Personalized Suggestions: Your AI sidekick will offer tailored insights based on captured memories. Automated Summaries and Action Items: Extracts key information from meetings, conversations, and class lectures. Searchable Memory Bank: Allows you to search your captured history and notes. Ask TwinMind Anything: Enables you to ask questions and receive answers with context of your memories or active meetings. Content Generation: Facilitates the creation of follow up emails, study guides, and other documents to help increase your productivity. Privacy Protection: Employs on-device transcription, choice to switch to private mode to store transcriptions locally, and no third-party data sharing. TwinMind aims to boost your productivity by providing a centralized, AI-powered platform that captures, organizes, and leverages your captured memories to offer personalized assistance and automate tasks, all while safeguarding your privacy.

Tools

fetch

ChatGPT

Retrieve a memory or task by known ID; searching first is optional. Provide exactly one of id or ids (up to 10 unique IDs). For summary-{meeting_id}, the text field contains the full currently available transcript in recording-time order and notes contains the latest user notes. This works while capture is in progress or its summary is pending, once content is available. Returns the full preferred transcript, using diarized text with speaker labels when available, even if live chunks are newer or longer; otherwise returns live transcript chunks. Timestamp prefixes are absolute times in ISO 8601 using the recorded transcript timezone when available, otherwise the local offset derived from summary local/UTC times (start time only, to whole seconds). If localization cannot be resolved, timestamps retain their original offset. Text without a known recording time stays untimed. Full means all available text from the preferred transcript source, including partial progress during capture; a live transcript trails the audio by about a minute, so re-fetch the same ID for newer text and notes. For todo-{todo_id}, returns task text and metadata. To retrieve a linked memory's content, fetch summary-{meeting_id} separately. Prefer prefixed IDs; raw IDs are also supported. A single id returns one item; ids returns a results array.

search

ChatGPT

Quickly search across the authenticated user's memories and tasks using one query. The query supplies keyword retrieval for both and semantic reranking for memory results; tasks use keyword relevance. Returns a results array containing the same memory summary objects as summary_search, including notes, actions, recording times, location, metadata, and fetch IDs. Task entries contain id, title, summary, and url. Memory IDs use summary-{meeting_id}; task IDs use todo-{todo_id}. Call fetch for a memory's full transcript and latest notes, or for task details. For separate keywords and a ranking question, or explicit date filters, use summary_search. Memory results may fall back to recent memories if keywords find nothing; check relevance. For the latest memories or a capture in progress, use summary_search without keywords or query instead; a capture without a summary yet may be absent, so fetch a known meeting_id directly.

summary_search

ChatGPT

Find the authenticated user's memories (recordings) by recency, recording date, or topic. For the latest memories, including a capture still in progress, omit both keywords and query: results are then newest first. Query reranks results by topic, which reorders a recent list and can push the newest memory out of it. An in-progress capture's summary reads "Still in progress"; fetch it for its partial transcript. A capture without a summary yet may be absent; fetch it directly if its meeting_id is known. For topics, pass keywords for retrieval and query to rerank by relevance. Use start_time/end_time for explicit date restrictions. If keywords find nothing, retrieval falls back to the most recent memories while preserving the time bounds. Returns a summaries array containing the same memory summary objects as search: id, title, summary, url, meeting_id, summary_id, meeting_title, notes, action, start_time_local, end_time_local, current_location, and metadata. The id can be passed directly to fetch. For the full transcript and latest notes, call fetch with id: summary-{meeting_id}.

todo_search

ChatGPT

Search or list the authenticated user's tasks. Results combine review tasks (AI suggestions awaiting confirmation) and main tasks (saved tasks), and identify each task's source. Use keywords to search task content and descriptions; omit keywords to list recent tasks. Completed and deleted tasks are excluded unless explicitly requested. Results include task details and IDs; fetch with id: todo-{id} retrieves a task. To read a linked memory's latest notes and full transcript, use its meeting_id in a separate fetch with id: summary-{meeting_id}.

fetch

Claude

search

Claude

summary_search

Claude

todo_search

Claude

Links

Documentation Privacy Policy

Similar Apps