Feature Ideas

To simplify workflow of working on artefacts in Claude based on the transcripted content of the HiDock P1. It would be wonderful to have an MCP access to these recordings in order to avoid manual text export and reimport in claude to continue the workflow

I want to pull my daily notes and pull them into my workflow tooling. Some may want to pull this into their AI Agent Framework like Claude Code, Claude CoWork, Amazon Quick, Kiro, Hermes Agent, OpenClaw, etc. The tools will change, but the source data export is too manual here. I don't consider this my unified AI Agent application, this is a tool to capture notes. These notes need to be exported automatically, or people will move to other tools that can be.

TODO items must be editable so users can correct or refine them when the automatically generated content is not accurate enough.

I already pay subscriptions for OpenAI, Claude & Gemini. Allow me to leverage those for my transcription vs having to buy credits from you.

Create a saved library of speaker names for quick selection. Ideally, this feature would work with voice recognition so Hinote can automatically match and assign the correct name from the library based on the speaker’s voice.

It would be great to offer an optional mind map-style summary, similar to what PLAUD AI provides. This format gives a clear and visual structure to the conversation—breaking down key topics, subtopics, and connections between ideas. For users dealing with complex calls or meetings, a mind map can improve understanding, memory retention, and even presentation to colleagues or clients.

I love this tool. BUT I would love to use this with my preferred AI for querying and using in various ways for work and personal life. I will pay more for this. I don't mind.

Allow user to add list of Domain-Specific Terminology, a mapping of the terminology and the explanation. So that when performing the transcript or the meeting note generation, this can be the input to increase the accuracy. I will be nice if import and export can be enabled as well.

For the HiNotes Version 3.0, please bring back the "Auto Transcribe". This was a very useful feature in HiNotes Version 1.0. I liked that when the audio clip is uploaded via the HiDock or an uploaded audio file, that it would automatically transcribe. I would prefer to look at the transcript rather than listen to audio or date/time to figure out which template I should select. Please have this as an optional feature that can be toggled on and off.

I use my HiDock P1 a lot, and I thinking if the folder navigation could be improved in a way where you can have sub folders for certain things. Example: [Class A] [topic 1] [Topic 2] Or Company 1 - legal meetings - marketing meetings - etc

Hi team, I’m using BlueCatch with my AirPods Pro to record calls on both iPhone and PC. Across both platforms, the real-time audio I hear during calls (monitoring/passthrough) is quite poor—speech is hard to understand and I often need to ask the other party to repeat themselves. However, the saved recordings are clear and intelligible on playback. If this discrepancy can be resolved via firmware or settings (e.g., monitoring path, codec/bitrate, gain/AGC, noise reduction), I’d really appreciate an improvement as soon as possible. As it stands, this otherwise great BlueCatch feature isn’t reliably usable for day-to-day calls on either platform. Expected Behavior Monitoring audio during the call should be comparable to the recorded audio so conversations remain intelligible without repeats. Actual Behavior Monitoring audio is muddy/unclear on both iPhone and PC, while the recordings are clean. Thanks in advance for looking into this!

One thing I noticed in the new Hi Notes UI, specifically in the Notes module, is that recorded calls must be added to folders manually. For example, I created a folder labeled “JUN1826,” but whenever I complete a call, I still need to manually add the recording to that folder. I am not sure whether this is possible, but it would be helpful if Hi Notes could automatically determine which folder a recorded call should be saved to. Perhaps there could be a “Condition Settings” feature that allows users to create rules in advance. This would enable recorded calls to be automatically organized based on predefined conditions instead of requiring users to move them manually. I normally receive around 8 to 10 calls per day, depending on the day, so this feature would save a significant amount of time. I understand that this may require additional development work, but it would be a very useful tool for end users.

What Was Built A 29-tool MCP server connecting Claude to HiNotes, reverse-engineered from the HiNotes web app since no public API exists. The server was built in TypeScript using the MCP SDK, with XML response parsing (HiNotes returns XML, not JSON), automatic session recovery, and packaged as a shareable Cowork plugin. Tools implemented across 7 categories: Notes, calendar events, todos, whispers, folders, templates (list/get/manage), vocabulary, and user/settings. Key technical challenges overcome: Auth requires both an Accesstoken header and a JSESSIONID cookie — discovered by reading browser DevTools Tokens are case-sensitive (one wrong character caused hours of debugging) Auto-retry on session_timeout: server calls /v1/user/info to get a fresh JSESSIONID, then retries the original request XML parser needed isArray hints to correctly handle multi-element responses Calendar response was triple-nested (data[] → events[] → events[] → calendarEvent) URL spaces must be %20 not + Several endpoints required browser JavaScript injection to discover correct methods/params How HiNotes Could Improve for MCP/AI Integration Publish an official API with documentation Everything here was reverse-engineered from browser traffic. An official REST or GraphQL API with docs would eliminate weeks of discovery work and make integrations reliable across app updates. 2. Offer API tokens with configurable expiry Current access tokens expire in a few days and can only be obtained via browser login with reCAPTCHA. Long-lived API keys (like GitHub's Personal Access Tokens) or OAuth 2.0 with a refresh token would make MCP integrations maintenance-free. reCAPTCHA on every token acquisition is the single biggest barrier to automation. 3. Add a search endpoint for notes There's no full-text search API. Finding notes by content requires fetching all notes and filtering client-side — expensive and incomplete. A GET /v1/note/search?q= endpoint would make Claude dramatically more useful for recall tasks. 4. Return JSON instead of XML XML parsing is fragile (tag nesting, array detection, attribute handling). JSON is the standard for modern APIs and would simplify every integration significantly. 5. Fix or document template/info POST /v1/template/info returns 400 for all parameter combinations tried. The template detail endpoint either doesn't exist or has undocumented requirements. As a result, Claude can list templates but can't retrieve their actual prompt content. 6. Expose note transcripts and summaries in the note detail endpoint GET /v1/note/recording/detail returns metadata only. Transcripts and AI summaries appear to be on separate endpoints not discovered during this project. Bundling these into the detail response (or documenting the endpoints) would make the most valuable HiNotes content accessible to AI. 7. Provide a native MCP server or webhook support The highest-leverage move: HiNotes could publish and maintain their own official MCP server, removing the maintenance burden from users entirely. Alternatively, webhooks on note creation/completion would enable real-time AI workflows.

I have the same people in meetings after but I have to name them each time. Otter ai has auto-speaker detection which is great. Would love to see this added.

In the “Recordings” tab, all files currently have default names like “20250604-161356-Rec87.hda,” which makes it hard to identify them quickly. It would be very helpful to allow users to rename a recording manually before summarizing it, directly from the interface. This small improvement would make it easier to organize and locate specific meetings, especially for users handling multiple recordings per day.