Meeting notes have quietly become one of the most contested battlegrounds in AI productivity software. For the past couple of years, a small wave of startups has convinced knowledge workers that they no longer need to scribble frantically during calls. Now Google is stepping into that space with a tool that does something most of those startups cannot: it runs entirely on your device, with no internet connection required.
The new app, called AI Edge Foresight, is a local-first meeting note-taker that transcribes conversations, generates structured notes, and answers follow-up questions using on-device AI. It is a direct shot at Granola, the well-regarded note-taking app that has built a loyal following among founders, journalists, and product teams. The difference in approach, though, is significant, and it says a lot about where Google thinks the next phase of AI competition will be fought.
What AI Edge Foresight Actually Does
At its core, AI Edge Foresight is a meeting companion. You open it before a call, it listens, and it produces a written record of what was said. That part is familiar territory. What sets it apart is that the transcription and note generation happen locally, meaning the audio and the resulting text do not need to travel to a remote server to be processed.
That distinction matters more than it might sound. Most cloud-based meeting assistants work by streaming audio to a data center, running it through a large language model, and sending the output back. It works well, but it means your conversation leaves your machine. For teams discussing product roadmaps, legal matters, or anything sensitive, that has always been an uncomfortable trade-off.
Google's approach flips the equation. The processing happens on the device itself, which means the app can function in airplane mode, on a spotty hotel Wi-Fi connection, or in a room where you would rather not have a microphone feed leaving the building. For a certain kind of user, that is not a nice-to-have feature. It is the entire reason to switch.
Why Local-First Is a Bigger Deal Than It Sounds
The phrase local-first AI has been circulating in developer circles for a while, but it is only now starting to reach mainstream productivity tools. The idea is simple: instead of treating the cloud as the default place for computation, you treat the device as the primary environment and use the cloud only when it genuinely adds value.
There are real trade-offs. On-device models are typically smaller and less capable than their cloud counterparts. They cannot draw on the same breadth of knowledge, and they may struggle with unusual accents, technical jargon, or overlapping speakers. But they are also faster, cheaper to run at scale, and far easier to justify from a compliance standpoint.
For companies operating under strict data rules, that last point is decisive. A tool that never sends audio off-device sidesteps a long list of legal reviews that cloud-based competitors have to endure. It is not a coincidence that Google is leaning into this framing. The company has been steadily positioning its Edge AI work as a privacy-forward alternative to the cloud-heavy approach that dominates the current market.
How It Stacks Up Against Granola
Granola built its reputation on a few things: a clean interface, thoughtful note formatting, and the ability to blend your own typed notes with AI-generated summaries. It has become a favorite among people who take meetings seriously and want a tool that feels like a writing surface rather than a surveillance device.
AI Edge Foresight enters that conversation with a different pitch. Where Granola emphasizes the quality of the note-taking experience, Google's app emphasizes the architecture underneath it. The question is whether users will care more about polish or privacy.
In practice, the two are not mutually exclusive, and the answer likely depends on the user. A solo founder on a podcast recording might not think twice about cloud processing. A lawyer on a client call almost certainly will. Google appears to be betting that the second group is larger, or at least more willing to switch tools over it.
Where Granola Still Has an Edge
Granola has had years to refine its note templates, integrations, and workflows. It plugs into calendar tools, syncs across devices, and has a community of users who have built habits around it. Switching costs are real, and a new entrant, even one from Google, has to offer something compelling enough to justify the move.
There is also the question of trust. Granola's entire brand is built on being a focused, independent tool. Google's brand is built on advertising, cloud services, and a long history of products that come and go. Some users will weigh that history heavily, regardless of how well the new app performs.
Where Google Pulls Ahead
On the other side of the ledger, Google has resources that no startup can match. It controls the underlying operating systems, the hardware, and in many cases the productivity suite that users already live in. If AI Edge Foresight integrates cleanly with Google's existing tools, the convenience factor could be substantial.
There is also the matter of scale. On-device AI improves as hardware improves, and Google ships a lot of hardware. Every new phone, laptop, and tablet with a capable neural processing unit is another device where an app like this can run well. That is a structural advantage that compounds over time.
The Broader Shift Toward On-Device AI
AI Edge Foresight is not an isolated product. It fits into a wider industry movement that has been accelerating over the past year. Apple, Qualcomm, Samsung, and a host of smaller players have all been pushing the idea that the most useful AI is the kind that runs locally, without a round trip to a server.
There are several reasons for this. Cloud inference is expensive at scale. Latency matters for real-time tasks like transcription. Privacy regulations are tightening in several major markets. And users, fatigued by subscription fees and data concerns, are increasingly receptive to tools that promise to keep their information close.
The catch is that on-device AI is harder to build. It requires optimization work that cloud-based tools can skip, and it demands hardware that not every user has. Google's decision to ship a local-first meeting app suggests the company believes the hardware is finally ready, and that the market is ready to reward the approach.
What This Means for the Meeting Notes Market
For the past two years, the meeting notes category has been defined by a handful of well-funded startups racing to add features. The next phase may look different. If local-first becomes a baseline expectation rather than a differentiator, the competition shifts to areas like accuracy, integration, and trust.
That is a harder game to win with marketing alone. It favors companies that can invest in model quality and hardware optimization, which is exactly the kind of game Google is built to play. Whether that translates into user adoption is a separate question, but the strategic logic is clear.
For users, the practical effect is likely positive. More competition in this space means better tools, clearer pricing, and more options for people who have been hesitant to hand their meeting audio to a cloud service. The arrival of on-device transcription as a mainstream feature is a meaningful step, not just a marketing angle.
Frequently Asked Questions
Does AI Edge Foresight work without an internet connection?
Yes, that is the central design goal. Because the transcription and note generation run on the device itself, the app can operate offline. This makes it usable on flights, in secure facilities, or anywhere the network is unreliable.
Is AI Edge Foresight a direct replacement for Granola?
It competes in the same category, but the two tools emphasize different things. Granola focuses on note quality and workflow integrations, while AI Edge Foresight leans on local processing and privacy. Whether one replaces the other depends on which of those priorities matters more to you.
What are the downsides of local-first meeting transcription?
On-device models are generally smaller than cloud-based ones, so accuracy can suffer with heavy accents, technical vocabulary, or several people speaking at once. Performance also depends on your hardware, which means older devices may not run the app as smoothly.
Will this push other companies toward on-device AI?
It is likely to accelerate an existing trend. Several major hardware and software companies have already been investing in local AI, and a high-profile entry from Google gives the approach more credibility with buyers and IT teams evaluating their options.
Who benefits most from a tool like this?
People who handle sensitive conversations, work in regulated industries, or frequently find themselves without reliable connectivity. For those users, the ability to keep meeting audio on the device is not a minor feature. It is the reason the tool exists.