Back to News
Tools

Instinct AI assistant gets its own email address to act on your behalf

Instinct AI assistant gets its own email address to act on your behalf
Image: TechCrunch

Instinct now assigns every user a dedicated inbox so its assistant can open accounts, email support lines and follow up on its own, pushing agentic AI deeper into everyday transactions.

September 9, 2026 · 2 min read · VintuxAI Newsroom

Instinct began giving its AI assistant its own email address on Tuesday, letting the tool contact businesses, register for accounts and chase customer support without the user copied on every message. The company is rolling the feature out to all users, free and paid, after the assistant built a following earlier this year on demos of booking flights and ordering groceries.

The mechanics are straightforward. Each user gets a unique Instinct address that the assistant draws on whenever a task needs written correspondence. Ask it to cancel a subscription or dispute a charge and it drafts and sends the emails itself, reads the replies, follows up when nothing comes back, and reports the outcome through its chat window. Paid tiers get higher sending limits and priority handling, while the free tier is capped.

What separates this from the browser-based agents that have dominated the category is the shift from clicking to corresponding. An email identity is a contact method that businesses already recognize, which lets Instinct work asynchronously — sending a request and waiting a day for a human to answer, rather than holding a session open. That format suits the boring administrative fights that agentic tools have struggled to finish, like refunds, cancellations and billing disputes, where the hard part is often persistence rather than navigation.

The autonomy cuts both ways, and security researchers have started picking at it. An assistant with a live inbox and the authority to write under a user's name is a plausible phishing target and a plausible source of accidental disclosure, since replies from third parties land in a mailbox the model reads. Instinct says it filters malicious senders and appends a clear AI disclosure to every outbound message. The company also says it will add approval gates for high-stakes actions in the coming weeks, though it has not said which actions will require a human sign-off.

Instinct has not disclosed adoption figures for the feature, which had a quiet early run before Tuesday's broader launch. The company is weighing expansion into SMS and messaging apps, according to people familiar with the plans, which would extend the same delegated-identity model to channels where AI disclosure is far less standard than in email.

The open question is accountability. When an assistant commits a user to a contract, misstates a cancellation date or leaks an account number in a thread, the paper trail points at a person who never typed the words. Instinct is betting that the saved effort outweighs that exposure, and the safeguards it ships next will be the first real test of whether users agree.

Written in-house by the VintuxAI newsroom, in our own words. The story was first reported by TechCrunch.