The Pulse
Hello Haven Launches AI Digital Twin With $15 Million Round
Hello Haven has emerged from stealth with a personal AI digital twin that stores user context in a private memory graph and performs tasks across phones, email, web and apps. The company also announced a $15 million pre-seed round led by Ma
AI.info Team ·
Hello Haven Wants an AI That Acts Like Its User
Hello Haven emerged from stealth on September 17 with a product that aims to do more than answer prompts. The company says Haven is a personal AI digital twin that builds a private memory graph for each user, learns preferences and past decisions, and completes tasks across phone calls, text messages, email, voice, the web and mobile applications.
The launch comes with a $15 million pre-seed funding round led by Mayfield. Hello Haven was founded by Frank Addante and Duc Chau, and the company says users in its private beta have interacted with their Havens more than 2 million times since the beta opened last month.
Haven’s central promise also creates the product’s main tension: the system is designed to act without requiring users to repeat their history, but it must make those decisions from a growing record of personal information and permissions. Hello Haven says that record belongs to and is controlled by the individual user.
From Personal Assistant to “Digital Twin”
Frank Addante, Haven’s founder and CEO, draws a clear distinction between the company’s product and a conventional assistant. “Haven is your true digital twin. While a personal assistant knows what you asked it to do, Haven knows what you would do and proactively does it for you,” Addante said in the launch announcement.
The company says Haven can prepare a daily briefing, triage an inbox, send follow-up messages, process returns, place orders and make calls. Users can also create recurring tasks in plain language, such as instructing Haven to send a reminder text or make a call before an appointment or meeting.
Those actions depend on the system retaining information about a user’s preferences, relationships, history, choices and boundaries. Hello Haven says two Havens can respond differently to the same request because each one reflects its individual user.
A Memory Layer Above the Models
Hello Haven describes Haven as a personal AI operating system rather than a single chatbot. Its orchestration layer is intended to hold a user’s identity, memory, permissions and intent, then select the model or combination of models and tools best suited to a task.
The company says its platform draws on more than 500 AI models and connects to more than 100 application integrations. The stated design keeps a user’s context in the orchestration layer instead of tying it to one model, allowing the same memory to carry between phone, voice, text, email, web and the Haven app.
That approach addresses a familiar weakness in general-purpose AI services: users often need to restate their history, constraints and preferences before receiving a useful answer. Hello Haven’s product is built around the opposite assumption—that the system should accumulate context and use it when deciding what to do next.
Mayfield Backs the Bet on Persistent Context
Mayfield managing partner Navin Chaddha said the company represents a move beyond chat interfaces toward software that remembers and performs work for a person.
“The first wave of AI gave us chatbots that could answer almost any question. The next wave will know you, remember what matters, and do the work. Personal AI that works persistently for you is the ultimate AI product,” Chaddha said.
The investment also reunites Mayfield with Addante, whom the firm previously backed at Rubicon Project. Hello Haven did not disclose a valuation or identify other participants in the funding round in its announcement.
Portability Is the Next Test
Hello Haven plans to open a developer platform where creators can build code and no-code applications that connect to a user’s Haven. The company also says it intends to make a digital twin portable across products and devices, including applications, televisions, automobiles, connected devices and robotics.
Duc Chau, Hello Haven’s co-founder, framed portability as a way to separate a person’s accumulated context from any single interface. “The interface will keep changing. A person’s context should not have to,” Chau said. “You’ll be able to sign in with Haven, and your memory, your permissions, and your judgment come with you.”
Those integrations are part of the company’s stated plan rather than capabilities fully described as available at launch. The product is available through the web at my.hellohaven.ai, as well as through Apple’s App Store and Google Play.
Private Memory, Delegated Action
Hello Haven says each user’s data is stored in a private personal memory graph owned and controlled solely by that individual. The company presents that structure as the foundation for persistent personalization and cross-service action.
The distinction matters because Haven is not limited to retrieving information or drafting responses. It is designed to send messages, make calls, place orders and carry out recurring instructions. The more authority users give the system, the more important its memory, permissions and interpretation of past decisions become.
Hello Haven is entering the market with a product that combines long-term personal memory, model selection and delegated action. Its first public test will be whether users trust a young service to remember enough to be useful while acting only within the boundaries they intended to set.