Former Plaud exec raises $26M for AI app that listens to you in the background

Ambi, a startup led by former Plaud executive Peter Mo, said on Monday that it has raised $26 million across multiple funding rounds. Luminous Ventures, which the company notes has backed Meituan and Unitree Robotics, and original design manufacturer Huaqin led the rounds. Mo is co-founder and CEO of the company, which is building an AI app designed to listen continuously and provide real-time intelligence and insights.
Alongside the funding announcement, Ambi is releasing apps for iPhones, Macs, and Apple Watch. The apps work like meeting notetakers, but users can choose to have them listen so they capture context throughout the day. Ambi's stated promise is that its AI agent can proactively offer suggestions or surface relevant knowledge while a meeting is happening, drawing on a user's calendar, meetings, knowledge base, and web data. Users can also begin a task on one device and pick it up on another.
The company is positioning this against the current generation of meeting notetaking tools. Those products generally focus on the tasks that follow a call, using summaries and action items gathered during the conversation plus connectors into other software systems. Ambi wants its agent to start the work during the meeting itself. That is a meaningful distinction for developers: a post-meeting summarizer can run as a batch job after the call ends, while an in-meeting agent has to operate on low-latency streaming input and decide in real time what is worth surfacing without becoming a distraction.
Ambi is not stopping at software. It plans a desktop app that will be able to look at the work users do and store important parts in its memory to give more personalized suggestions. It is also planning its own hardware device with a screen, a speaker, a camera, and microphones to capture context and display information. That combination, continuous ambient capture plus a persistent memory, is the same broad approach behind products such as Littlebird and Minimi. The practical challenge for all of them is deciding what to retain, since the more context an agent stores, the better its suggestions can be, but also the larger the surface for privacy mistakes and the harder it becomes to keep retrieval useful rather than noisy.
Privacy is the open question hanging over the launch. There is growing concern about AI devices that lack sufficient indicators that they are capturing data about people around the user. It is not clear whether Ambi's apps or its upcoming device will have enough indicators to alert third parties that their conversations are being recorded if the user does not disclose their use of the technology. Some meeting notetakers have been sued over recording people's conversations and data without consent, and the recent release of Apple's Audio Intelligence feature, which transcribes speech continuously, has raised similar privacy concerns.
Mo said users can turn off ambient data capture on Ambi's apps and devices at any time, and that the feature is off by default. The startup said it does not store recordings of users and that it encrypts data, though it did not specify what kind of encryption it uses. Even with those measures, the responsibility for declaring ambient capture falls on users. Mo described this as an industry-wide problem, saying the company tells users they must inform others about ambient capture and that the industry has not solved disclosure, adding that if users try to hide the capture, the company cannot solve that problem. He said Ambi has tried to put information about disclosure in every place possible.
Ambi currently has more than 50 staff across offices in San Francisco, Beijing, Shanghai, and Shenzhen. The company is hiring for its go-to-market function and for AI engineers to work on speech-recognition models. Those roles point at where the difficult engineering sits: running speech recognition continuously, cheaply, and accurately enough to be useful, on battery-powered devices, is a heavy lift, and in-house models can reduce dependence on third-party APIs for audio that is inherently sensitive.
The company did not disclose the individual round sizes, valuations, or how the $26 million will be allocated.
Why it matters: Developers and product teams building meeting assistants or ambient agents now have a well-funded competitor whose bet is that in-meeting, proactive assistance beats post-meeting summarization. The funding also signals continued investor appetite for always-on context capture despite unresolved consent questions, which likely keeps pressure on teams to make recording indicators, default-off settings, and data retention policies first-class design decisions rather than afterthoughts. Inference: expect procurement and platform review at workplaces to focus on whether an assistant can prove it does not retain recordings and can be turned off verifiably.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
One of the core promises of AI is that the more it knows about a person, the better it will be able to assist them with tasks and make appropriate suggestions, which, if you'll believe the tech companies, requires apps and devices to listen, watch, or record users' activity all the time and gather all possible context.