todo.is

People have errands and small jobs (booking a table, comparing flights, summarizing invoices, monitoring a competitor's pricing page) that require opening a browser, filling forms, or digging through email, which is tedious and interrupts their day. Existing automation tools like IFTTT or Zapier require building a flow with triggers, actions and field mapping, which only works for fixed pipelines and breaks when a site changes. These tools also can't handle one-off, messy, or judgment-based tasks, or sites without an integration. The result is that people either do these small jobs manually or don't do them at all.

todo.is gives each user an AI agent, created in seconds, that lives inside chat apps they already use: web chat, WhatsApp, Telegram, Slack, Discord, email, Feishu/Lark, DingTalk, Matrix, QQ and WeCom. Users write a to-do in plain language (e.g. 'book a table for 4 on Friday at 8' or 'every morning at 8, summarize my inbox'), and the agent plans the steps, browses real websites with an actual browser, runs code, handles files, and produces text, images, video or voice notes as output. It asks for explicit approval before sensitive actions like sending emails, making purchases, changing calendars or deleting files, and logs every step in an Activity timeline. Each agent has its own sandboxed workspace, memory the user can view and edit, runs 24/7 in the cloud, and supports either platform AI credits or a user's own API key across 450+ models.