Granola, an AI-powered note-taking app, is worth $250 million and has become a popular tool among tech founders and venture capitalists. But one developer believes there is a need for a more private, local-only alternative that is available for a one-time fee without a subscription. That led to the creation of a new Mac app called Talat.
Nick Payne, a developer based in Yorkshire, UK and a self-proclaimed computer geek, says the idea to build a local AI note-taker came about largely through a series of happy coincidences.
“I think Granola is great. It’s a shining example of what you can do with Electron apps (a framework for building desktop applications) if you give it enough love and care,” he told TechCrunch. “When I first tried it, I was fascinated by the ability to record system audio on a Mac without recording video, which was a standard workaround at the time. This led to a ton of research and the discovery of a relatively new and poorly documented Apple API.”
To make it easier to work with its API (Core Audio Taps, which lets developers take advantage of the Mac’s audio streams), Payne decided to create AudioTee, an open-source audio library.
“During that time, we were slowly putting together a toolkit, but we couldn’t find anything that felt like it could stand on its own as a product and not just a cool tech demo,” Payne said. “The state-of-the-art hosted transcription model, the same provider that people like Granola use, is incredible, and it’s intuitively great to see your speech unfold on the screen in near real time. But the tradeoff was always that I needed to provide not just my data, but my audio data, my actual voice,” he added.
He then came across a software toolkit called FluidAudio, a Swift framework that enables completely local, low-latency audio AI on Apple devices. This allows you to run small, fast transcription models directly on Mac’s Neural Engine, Apple’s dedicated AI processing hardware.
This was the work that made Payne realize that he could turn his research into an actual product. No audio leaves your Mac, and no transcripts are stored on third-party servers.
Talat was built with Payne’s long-time friend and former colleague Mike Franklin, and is the result of Payne’s interest in the audio field. The result is a one-time purchase of 20 MB without the need to create an account or share analytics data with developers. There are no ongoing fees.
While some AI note takers may have more bells and whistles, Talat offers a streamlined set of features. Capture audio from your computer’s microphone while using meeting apps like Zoom, Teams, and Meet and transcribe it in real time. The app attempts to assign speakers in real time, but you can reassign them if needed. You can also take notes and edit, delete, and split transcript segments. At the end of the meeting, your local LLM will generate a summary with key points, decisions, and action items.
Notes, transcripts, and summaries are also all searchable on Talat.
In addition to privacy considerations, Payne said the goal is to give users more choice.
“We’re all about configurability and giving users control over where their data goes: choosing their own LLM, auto-exporting to[note-taking app]Obsidian, webhooks to push data at the end of a meeting, and an MCP server. This is a standardized way for AI tools to connect to external data sources and ‘pull on demand,’” he explained.
Under the hood, the AI is a mixture, “mostly stitched together and abstracted behind FluidAudio,” Payne said, which he believes does a lot of the heavy lifting. To summarize, the app defaults to an Al model called Qwen3-4B-4bit and can run on fairly modest hardware.
However, users can also switch this to any cloud LLM provider. You can also choose between two variants of Parakeet (a speech recognition model developed by Nvidia) or Ollama (a tool for running AI models locally), giving you more control over the experience. Over time, Talat plans to add support for more built-in choices and integrate with other apps like Google Calendar and Notion.
At launch, users with M-series Mac computers (computers with Apple’s own processors starting with M1) will be able to download the app and try 10 hours of recording for free before making a purchase decision.
Talat is available for $49 in this pre-release version, which is currently in development.
Once the app reaches the 1.0 release, the price will increase to $99.
Payne and Franklin started Talat in-house and plan to continue making their core product a one-time purchase.
