There are already plenty of products that will join a video call, record it, and generate a summary. Building another one only makes sense if it begins with a different set of assumptions.
Quendian began with a simple question: what would meeting transcription look like if the user stayed in control?
That question changes almost every product decision. It changes how the app is distributed, how it connects to AI providers, where the finished notes go, and which meetings it can record.
Start with a Mac app, not another meeting participant
Many meeting assistants work by inviting a bot into a call. That can be useful, but it also makes the transcription service part of the meeting itself. The bot appears in the participant list, depends on an integration with the meeting platform, and may not work when the conversation happens somewhere else.
Quendian takes a different approach. It is a native macOS app that records the microphone and the audio playing through the Mac when the user starts a recording. It does not need to join the call as a participant.
That means the same workflow can work with Zoom, Google Meet, Microsoft Teams, a browser-based interview tool, a webinar, or another source of conversation on the Mac. Quendian is not tied to the roadmap or API of one meeting platform.
The user still has the responsibility to record appropriately and obtain consent where required. The important product distinction is that the recording starts with an intentional action on their own device, not with an invisible integration operating somewhere else.
Distribution should contribute to trust
Quendian is being built for distribution through the Mac App Store.
App Store review is not a guarantee that software can never contain a bug or make a mistake. It does, however, add meaningful constraints and an independent checkpoint. The app must be signed, use Apple’s distribution process, describe its behavior, work within the macOS sandbox, and justify the permissions it requests.
Those constraints are useful for an application that can access a microphone and system audio. They force privacy and permission handling to be part of the architecture rather than a paragraph added at launch.
The App Store also gives people a familiar installation and update path. They do not have to decide whether to trust an unsigned download from an unfamiliar website before they can evaluate the product.
Bring the API keys you already trust
The current AI market changes quickly. Models improve, pricing shifts, and different providers are better at different jobs. A meeting app should not require users to surrender that choice.
Quendian supports bring-your-own-key configurations for transcription and summarization. A user can connect supported providers directly, use compatible local services such as Ollama or LM Studio, and decide where processing happens.
In bring-your-own-key mode, Quendian does not proxy the request through a Quendian server. The credential stays on the Mac, and the audio or transcript goes directly to the provider the user selected.
Quendian Cloud exists for people who prefer convenience over provider setup, but it is an option rather than a prerequisite. The product should be useful without creating an account or committing to one hosted AI stack.
Notes should leave the app
A transcript is valuable because of what happens after the meeting. It may become a project update, a set of action items, research material, a client record, or context for another tool.
That value should not be trapped inside the application that generated it.
Quendian writes completed sessions as Markdown files to a folder the user chooses. Markdown is intentionally ordinary. It can be opened in a text editor, indexed by Spotlight, committed to a repository, copied into a knowledge base, synced with a storage provider, or used by another application.
The app also keeps its working state in its sandbox so the interface remains fast and reliable, but the durable user-facing output is portable. If someone stops using Quendian, their notes do not disappear behind an export screen or an expired subscription.
Control is a product feature
None of these choices produces the shortest possible feature list. Supporting different providers is more work than selecting one. Portable files require careful handling of folder permissions and updates. Recording system audio across meeting tools is more complex than building for a single integration. App Store distribution adds review and sandbox requirements.
But those constraints create the product I wanted to use:
- one recorder for the conversations happening on the Mac;
- provider choice instead of a mandatory AI backend;
- local and portable notes instead of another closed workspace; and
- a familiar, reviewed distribution channel for software with sensitive permissions.
That is why build another meeting transcript service. Not because transcription is unsolved, but because ownership, portability, and choice still deserve a better default.