Caddy Wants to Be the Proactive AI Assistant That Texts You First
Most “AI assistants” today still wait for you to ask. Caddy, a Y Combinator Fall 2025 startup founded by alumni of Loom’s AI team, is taking a different approach: an AI personal assistant app that lives in your Messages, reaches out when something needs your attention, and gets work done across the tools you already use.
Founded in 2025 by Connor Waslo and Rajiv Sancheti, Caddy bills itself as a personal AI that texts you first. Rather than waiting to be prompted, Caddy monitors your connected apps and sends you a message when there’s something worth knowing—an important email that needs a reply, a meeting that needs to be added to your calendar, a follow-up that slipped through the cracks. The platform integrates with the everyday stack of modern knowledge workers: email, calendar, Slack, Linear, Notion, and more. The thesis is that the most useful AI is one that meets you where you already are – in your messages – instead of expecting you to remember to check yet another interface.
The product covers a wide range of workflows. Caddy can replace your reminders app by turning natural-language texts into reminders for errands, deadlines, and recurring tasks. It monitors your inbox and alerts you only to the emails that warrant immediate attention. It helps keep your calendar current by turning plans and obligations mentioned in conversation into events. It can track expenses from receipt screenshots, draft email replies in your voice, coordinate scheduling, surface action items from meeting notes, and deliver a short morning brief with what actually matters for the day. The thread running through all of it is the same: Caddy texts first when something matters, so you don’t have to constantly check.
The founders bring an unusually strong product pedigree to the problem. Waslo previously led product for Loom’s AI Suite and the company’s Monetization team. Sancheti led design for Loom’s AI suite and earlier worked at Airbnb and as a KP Fellow. Both watched the rise of ambient AI from inside one of the most successful asynchronous-work products of the past decade, and Caddy is a direct extension of what they learned: context is everything, and the difference between a useful AI and a forgettable one is whether it actually finishes the work.
The market backdrop is favorable. Knowledge workers spend a growing share of their day in coordination overhead – drafting messages, updating tickets, summarizing meetings, and rerouting follow-ups across half a dozen tools. The first wave of LLM products tried to reduce that overhead by offering chat interfaces inside each tool. Caddy’s bet is that the second wave will be agents that live above those tools, hold long-term context about a person’s responsibilities, and reach out proactively when action is needed – meeting users in the interface they already have open all day: their messages.
With a tight team and a clear opinion about what proactive AI should feel like, Caddy is one of the more interesting entries in the increasingly crowded “AI agent for knowledge work” category. By routing everything through Messages rather than building yet another app to check, Caddy is making a pointed bet: the best AI personal assistant app is the one that fits into your life, not the one that asks your life to fit around it.