What is flowflow?

FlowFlow is an iOS voice notes app with a twist: instead of just storing recordings, it lets you have a conversation with your notes afterward. The idea is that when an idea strikes while you are walking or in between tasks, you can quickly record it, and later ask questions like what was that pricing idea I mentioned, and the app finds the relevant passages and links back to the original recordings. When you record a note, the audio is automatically transcribed into text, filler words like um and hmm are stripped out, and the text is converted into a numerical representation called an embedding that captures its meaning. These embeddings are stored in a local vector database on your device, a specialized tool for searching by meaning rather than by exact keywords. When you later ask a question, the same process happens to your question, and the app finds notes that are semantically similar, then uses an AI language model to compose a proper answer with links to the sources. The AI chat can also take actions on its own, searching notes, creating new notes, or summarizing folders, through what are called agent tools, chaining up to four tool calls before answering. You can also import documents in formats like PDF, Word documents, text files, and CSV to include them in the searchable knowledge base, with automatic OCR for scanned PDFs. The entire app is written in Rust, a systems programming language known for performance and safety, and uses Dioxus, a Rust framework for building native iOS user interfaces. All data stays on your device, since it uses a local database, and nothing is sent externally except the AI model API calls for embeddings, transcription, and chat, where you can choose between OpenAI and Anthropic. It is licensed under EUPL v1.2.