Tab AI assistant Emerges From Stealth at $300M

A new personal AI assistant wants to live where you already spend your time: your text messages. San Francisco startup Tab emerged from stealth this week with a reported $300 million valuation, betting that the next great AI assistant won't be an app you open — it will be a contact you text.

The premise of the Tab AI assistant is deliberately simple. Instead of a separate dashboard or a voice speaker, users text Tab on iMessage or WhatsApp the way they would message a person they trust. The agent then completes real tasks: booking flights, ordering groceries, calling businesses, drafting emails, and chasing wire transfers. It acts rather than merely advising.

According to reporting by TechCrunch, Tab announced the launch on Tuesday, October 6, and declined to disclose exactly how much it raised. Investors in the round include SV Angel, Valar Ventures, and American Spirit. The company is growing mostly by word of mouth for now and says a new product is on the way soon.

Why texting won the interface race

Tab's founders argue that the battle for the consumer AI assistant will not be won by the flashiest demo. Brennan Erbz, Stafford Schlitt, and Ammar Amdani started building the service earlier this year around the observation that people already manage their lives inside messaging apps. Texting Tab, Amdani told TechCrunch, should feel like asking someone reliable to take the weight off — the bills, the bookings, the phone calls people keep putting off, and remembering the people they love.

The technical setup is what makes the Tab AI assistant different from a chatbot with a phone number. Tab says each user gets an agent with its own phone number, its own computer, and its own wallet, so it has the tools to execute tasks on the user's behalf rather than handing back instructions. That design puts it squarely in a crowded new category of consumer agents that handle everyday chores — a trend that has recently drawn heavyweight backing, including Meta's push into personal agents and commerce-focused protocols from companies like Sierra. GenZai previously covered the Sierra and Meta personal agent protocol for AI commerce, and the race toward an agent that buys and books for you is also visible in TikTok's new AI shopping assistant.

That positioning matters because the market is moving fast. Meta's own personal agent made a splash just last week, and the broader pattern is clear: assistants are shifting from answering questions to taking actions across email, calendars, phones, and payment systems.

Trust is the real product

An assistant that can read email, make phone calls, and spend money is only as good as the trust behind it — a point Amdani himself raised with TechCrunch. That framing is strategic as much as it is honest: the permission model is the make-or-break question for every agent that touches real accounts and real wallets.

Tab's stated safeguards are designed to answer that question up front. The company says user data will not be used to train its AI models. Payment card details are stored in a vault that Tab itself cannot access, and no irreversible action is taken without the user's explicit approval, according to the launch coverage. Access is currently through a waitlist, with no disclosed user counts or revenue figures.

Independent verification of those claims will take time. The founders have not detailed how the vault or the approval step works under the hood, and no third-party testing has been published yet. For now, the guarantees are the company's own claims relayed in interviews, and early adopters are effectively trusting Tab's word on data handling and purchase safeguards.

Why it matters

The Tab AI assistant is the latest signal that the AI assistant wars have entered a new phase: execution over conversation. Investors handing a $300 million valuation to a pre-seed startup that barely existed publicly last month says the appetite for agent infrastructure is still enormous. If Tab can prove that people will hand real errands — and real payment credentials — to a text thread, the template could reshape how everyday services get used. And with agent race costs still dominated by ever-cheaper, ever-faster models like the ones GenZai covered in the Claude Haiku 5.5 price-cut story, the economics of giving every user their own personal agent are moving fast in Tab's favor. The open question, as always, is trust: whether users actually let it in.