The news
Personal AI assistants that take actions for users, rather than just answering questions, gained three new entrants on October 6 and 7, 2026, and each one led with privacy. Tab came out of stealth on October 7 at a $300 million valuation, TechCrunch reported, joining Underdog and Hark, which both launched on October 6.
Tab, founded by Brennan Erbz, Stafford Schlitt and Ammar Amdani, works over iMessage or WhatsApp and handles errands such as ordering groceries, making reservations and booking travel, according to TechCrunch. The assistant has its own phone number, computer and wallet. Investors include SV Angel, Valar Ventures and American Spirit; the size of the round was not disclosed. Tab told TechCrunch that user data is not used to train models, that the payment card sits in a vault Tab cannot access and that it takes no action without explicit approval.
Underdog, from Conway Research and founder Sigil Wen, takes a different route: it runs entirely on the user's own Mac or Windows PC, with Linux, iPhone and Android versions said to be coming, TechCrunch reported. It uses a 27-billion-parameter reasoning model and encrypts the keys that connect it to email and other accounts. Backers include Andreessen Horowitz, Khosla Ventures and SV Angel. Underdog is free, will not carry ads and plans to earn a small cut of payments processed through Stripe. It launched as an invite-only beta.
Hark, founded by Brett Adcock, released its assistant widely on October 6 with a free tier and a paid subscription for heavy users, TechCrunch reported. Hark can access a user's email, calendar, files and credit cards, suggests tasks such as approving expenses or booking flights, and shows its web browsing in a small window so users can watch what it does. The company said it will not sell ads or user data and plans an AI-native device in 2027.
All three pitch themselves against Instinct, which TechCrunch noted raised a $1 billion Series C at a $10 billion valuation, and Meta's Muse.
The numbers
- Tab valuation
- $300 million
- Underdog model size
- 27 billion parameters
- Instinct Series C (for comparison)
- $1 billion at a $10 billion valuation
Why CEOs should care
For CISOs, these products are the next shadow IT problem. Hark asks for email, calendar, files and credit cards; Tab texts with users on personal messaging apps. Employees who connect a work inbox or corporate card to a consumer assistant move company data outside approved systems. Update acceptable-use policies to name AI agents explicitly, and check whether identity tools can see OAuth grants to these apps.
For buyers evaluating agents, the three launches give a useful checklist. Ask where data is processed (on the device, as with Underdog, or in the vendor's cloud), whether it trains models, who can access stored payment credentials, and whether each action needs approval. Tab's vault and Hark's visible browsing window are the kinds of controls enterprise versions should match.
For CFOs and boards, the business models differ in ways that affect risk. Underdog says it will earn from payment interchange rather than ads; Hark sells subscriptions. A vendor's revenue model tells you what it is paid to optimize, so ask before employees start using these tools for expense approvals or purchases.
The bigger picture
The rush follows big money at the top of the market, including Instinct's $10 billion valuation. With capable models widely available, startups are trying to stand apart on trust and data handling rather than raw capability, a sign privacy is becoming a buying criterion for agents that touch personal and work accounts.
What’s next
Watch whether any of the three adds business accounts with admin controls and audit logs, how Underdog's invite-only beta expands to phones, and whether Hark's planned 2027 device arrives on schedule.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
