The personal AI race is moving off the screen. Two new assistants, Ambi and Tab, are betting that software can become more useful by taking on tasks, remembering context, and staying close to users throughout the day. That promise arrives with a familiar complication: the more an assistant knows, the more users must trust it.
Peter Mo, co-founder and CEO of Ambi and a former Plaud executive, announced on October 12, 2026, that the company had raised $26 million across multiple funding rounds. Luminous Ventures led the funding, with support from Original Design Manufacturer Huaqin; Luminous has backed companies including Meituan and Unitree Robotics.
Ambi is releasing apps for iPhones, Macs, and Apple Watch. The apps act as meeting note-takers that can listen and capture context all day, giving the assistant information it can use beyond a single conversation.
“The way we interact with intelligence is limited. Often the intelligence is trapped on the screen, and if you use hardware like Plaud, you need to wait for recording to transfer to the phone, upload to the cloud, and get intelligence after processing. That’s why we are building Ambi to use AI anytime and anywhere,” Mo said.
Ambi wants context without making users ask
Ambi’s AI agent can surface relevant knowledge or offer suggestions during meetings by drawing on calendar data, meetings, a knowledge base, and web data. Users can start a task on one device and continue it on another, turning the assistant into a running workspace rather than another isolated app.
An upcoming desktop app will look at a user’s work and store important parts in its memory. Ambi plans to use that information for personalized suggestions, which sounds useful until the phrase “look at your work” meets the privacy policy.
The company also plans its own hardware device with a screen, speaker, camera, and microphones. Ambi has over 50 staff across offices in San Francisco, Beijing, Shanghai, and Shenzhen, giving the company a substantial team for a product that still needs to solve the social rules around ambient listening.
Users can turn off ambient data capture on Ambi’s apps and devices, and the feature is off by default. Ambi says it does not store recordings of users and encrypts data, but it has not specified the encryption method.
Ambi also says users must inform other people when ambient capture is active. That requirement leaves the hardest part unresolved: the people being recorded may not know what the device is doing, and the industry has not solved that disclosure problem.
Tab sells delegated work through trust
Tab took a different route when it announced its emergence from stealth on October 7, 2026, with a $300 million valuation. The company declined to share exact funding details but said it raised money from SV Angel, Valar Ventures, and American Spirit.
Users interact with Tab by texting the assistant through iMessage or WhatsApp. Tab can order groceries or find presents, and it does not explain how to complete those tasks—it completes them for the user.
The assistant has its own phone number, computer, and wallet. Tab is growing mostly through word of mouth and plans to launch a new product soon, suggesting that the company sees personal referrals as a stronger sales pitch than another parade of feature demos.
Tab’s central argument is trust. The company says data is never used to train AI, while card information stays in a vault that Tab itself never sees.
“Your data is never used to train AI. Your card lives in a vault Tab itself never sees. Nothing that can be undone happens without your ‘yes.’ Capability is table stakes. Trust is the product,” said Ammar Amdani.
That line captures the divide between the two products. Ambi wants to collect more context so it can anticipate needs across devices, meetings, and work; Tab wants users to hand over tasks without worrying about what happens behind the scenes.
Tab’s capability is considered less important than trust, while Ambi’s appeal depends on constant access to personal context. Both companies are selling assistants that act beyond a chat window—but the winning feature may be knowing exactly what the assistant remembers, when it is listening, and who gets to decide.
Based on
