San Francisco-based Nirva has raised $25 million in pre-Series A funding to launch an AI-powered wearable designed to continuously analyze its owner's speech and provide personalized insights.
The round follows $8 million raised earlier in 2026 from investors including Lightspeed Venture Partners and South Park Commons, bringing total financing to $33 million.
Founded in 2025 by Wei Lyu, Jason Chen and Hanying Hu, Nirva combines experience in consumer hardware, augmented reality and jewelry design.
Lyu and Chen previously worked at Meta Reality Labs, while Hu founded a fashion and jewelry business.
The company's first product is a lightweight AI pendant that can also be worn as a bracelet.
It is designed to look like conventional jewelry rather than a visible recording device, with a focus on making AI technology acceptable for everyday wear.
The device weighs approximately eight grams and is expected to provide up to three days of battery life.
Nirva plans to open orders in November 2026 at a price of $249, with initial shipments targeted for the end of the year.
The device will be available in several finishes, including silver, black and gold.
Its central feature is a personalized voice recognition system.
During setup, users record samples of their voice, allowing Nirva to develop a model intended to distinguish the wearer from other speakers.
The device then continuously listens for the owner's speech rather than deliberately recording complete conversations involving surrounding people.
Nirva uses the information to build a personalized record of daily experiences, communication patterns and social interactions.
Its software analyzes factors including speaking speed, tone, pitch and volume to derive insights about possible emotional patterns and communication habits.
Rather than functioning primarily as a transcription tool, Nirva is attempting to create a system that understands a user's behavior over time.
For example, it might identify recurring communication patterns, remind the user about something mentioned earlier or suggest ways to improve interpersonal interactions.
The company calls its underlying personalization system Nirva Core, which uses existing frontier AI models alongside its own processing and memory capabilities.
Nirva says audio recordings are discarded after processing, while relevant information is retained to support personalized insights.
Those privacy protections are particularly important because the product is designed to operate throughout the day.
The company reportedly has approximately 9,000 people on its waiting list ahead of launch.
Nirva is entering a growing market for AI wearables, alongside products from companies such as Friend, Amazon's Bee and other startups developing ambient computing devices.
These businesses share a broader thesis: AI becomes more useful when it has access to ongoing personal context rather than relying exclusively on information manually entered into a chatbot.
However, the market is still searching for a compelling consumer use case.
Users must be willing to wear the hardware regularly, keep it charged and trust that the information it collects will remain private.
Nirva's voice recognition approach may provide some differentiation, but recognizing a wearer's speech does not eliminate every privacy concern associated with an always-listening microphone.
The company will need to demonstrate how it handles background voices, sensitive information, consent and data retention in practice.
Its consumer value proposition also remains unproven.
AI-generated emotional and behavioral insights may interest early adopters, but maintaining long-term engagement will require the product to deliver genuinely useful recommendations without becoming repetitive or intrusive.
Hardware margins, manufacturing quality and customer support will add further operational challenges.
Nirva's next milestone is its November order launch, followed by initial deliveries and evidence of sustained customer usage.
The broader question is whether AI wearables can become a daily computing category or remain niche devices for early adopters interested in personal analytics.