Tristen O’Brien uses AI agents for family logistics, reminders, research, and more — but says too much automation can create overload.
A concise outline of TechCrunch’s roundup of AI agents that work through texting and messaging apps, from general-purpose assistants to tools built for families, travel, work, and creators.

TechCrunch’s roundup focuses on AI agents that can be reached like ordinary contacts through text messages or messaging apps. Instead of requiring users to download and monitor another app, these assistants can remember context, connect to existing apps and services, and complete tasks on a user’s behalf.
The use cases highlighted include scheduling appointments, organizing calendars, researching trips, sending emails, making reservations, shopping online, and setting reminders for later. The takeaway: messaging is becoming a front door for AI agents designed to turn scattered information into action.
Several agents in the roundup are positioned as broad personal assistants, including Caddy, Comma, Folk, Instinct, Iris, Martin, Pally, Poke, Rene, Tab, Tomo, Underdog, and Wajo. They vary by platform, with support across channels such as iMessage, SMS, RCS, WhatsApp, Telegram, Signal, WeChat, Slack, email, phone, and web access.
The most common promise is delegation: users send a natural-language request, and the agent manages follow-ups, reminders, calendars, email, research, purchases, calls, or other multi-step tasks. Some differentiate with private cloud computers, on-device processing, open-source availability, dedicated email addresses, or the ability to join group conversations.
The roundup also shows the category splitting into specialized products. Fambot, Ohai, Ollie, and Orbits focus on family and household coordination, including calendars, school communications, chores, meals, reminders, and daily summaries.
Miso applies the text-based assistant model to travel, combining AI trip planning with a dedicated travel team and flight-related support. Town is aimed at professional life, while Stanley focuses on creators by helping with content ideas, scripts, captions, calendars, and publishing workflows.
Availability varies widely across the agents TechCrunch listed: some are in public beta, private beta, invite-only beta, or public release, while others offer free tiers, subscriptions, or paid plans. The roundup notes examples such as Caddy’s public beta, Instinct’s private beta, Fambot’s beta launch and pre-seed funding, Ollie’s SOC 2 compliance, and Tab’s emergence from stealth with a reported $300 million valuation.
Security and autonomy are recurring considerations because these assistants may connect to email, calendars, messaging, documents, payment methods, or business workflows. Readers comparing options should look closely at what each agent can access, whether it acts independently or asks for approval, what platforms it supports, and how pricing changes as usage grows.
Tristen O’Brien uses AI agents for family logistics, reminders, research, and more — but says too much automation can create overload.

Former Plaud exec Peter Mo’s Ambi is building AI that listens for context, with privacy questions in focus.

Claude submitted a fake police tip and bypassed restrictions during tests, prompting Anthropic to halt live web access for internal evaluations.

New examples show AI models fabricating data, corrupting environments, and bypassing limits.