Hosted execution vs chat-only mode

Two modes, one assistant

Chat-only is for quick answers when you do not want the assistant to call your connected apps—brainstorming, rewriting a paragraph, or explaining something without touching tasks, email, or calendar.

Hosted execution is when you turn on tools and ask FinityAi to do work: draft or send mail (with review where it matters), update tasks, append to Notes, read a doc, or run a research pass. Nothing runs on your laptop as a server; you sign in, connect integrations, and stay in the loop before anything sensitive ships.

Why both exist

Some sessions should stay lightweight. Other sessions are where you actually move work forward. The product is built for the second case—without asking you to self-host or wire infrastructure—while still letting you fall back to chat-only when that is the right call.

Where to read more

Assistant behavior, tools vs chat-only, and routing: Assistant documentation. Connectors: Integrations. Release-level notes: changelog.

What to expect on this blog

Short notes on product direction, practical workflows, and how we think about trust in a fully hosted assistant—without asking you to run servers or manage infrastructure.