Features

A real assistant, built on your own AI.

Dorinda is the accountability layer for the ChatGPT or Claude you already use — it makes your AI actually follow through, with the memory and controls to back it up. What you decide becomes a tracked responsibility, not a message that disappears — for you, or shared with your household. No new app to download, learn, and abandon; it lives right inside the assistant you've got.

  • Your data never trains a model
  • No server-side AI — a deterministic record
  • You own it and can export anytime
Lives where you already are

No new app to download and abandon — Dorinda plugs into the ChatGPT and Claude you already use.

You connect the ChatGPT or Claude you already pay for over MCP, and Dorinda becomes the accountability layer it works through — turning it into an expert personal assistant grounded in your real record, with the memory and controls to back it up. Same AI, nothing new to learn, no lock-in.

So it's not one more app you download, use for three days, and forget.

  • Bring the AI you already pay for — no lock-in.
  • A dependable record it can trust — the same facts every time.
  • Your data is never used to train a model.
Delegations & reminders

The unit of work is a responsibility, not a task.

In plain terms: you tell it what you need, and it stays on that until it's actually done.

A Delegation is a durable, closed-loop responsibility — tracked until it is actually done, with who owes what, what is waiting, and what is at risk. It does not vanish when the chat scrolls; it stays open, reminding you until the loop closes.

So the thing you meant to do doesn't vanish the second the chat scrolls.

  • A clear “done when,” so completion is unambiguous.
  • Ownership and waiting-on tracked on every item.
  • Persistent reminders and follow-through until it is closed.
Follows up & waiting-on

It chases the people you're waiting on.

Waiting on the contractor, the doctor's office, a recruiter? Dorinda tracks exactly what you're waiting on and from whom, keeps it visible instead of forgotten, and — with your OK — nudges the person so the reply doesn't stall out.

So a reply you're owed stops silently rotting in someone else's inbox.

  • See what you're waiting on, and from whom, at a glance.
  • Dorinda drafts the nudge; it goes out only on your approval.
  • A stalled reply becomes visible, not forgotten.
Governance & approvals

High-risk actions always wait for your yes.

You set the rules. Sending an external message, spending money, or contacting someone new always requires your approval — right in the chat. Low-risk actions can be auto-allowed. The AI can prepare and stage; it cannot act past your line on its own.

So nothing goes out in your name until you've said yes.

  • External sends, spend, and new contacts are gated.
  • Approve or decline inline, in the conversation.
  • Auto-allow the low-risk, so it stays out of your way.
Connected email & calendar

Act from your own accounts — with your permission.

Connect your Gmail or Outlook (Microsoft) mail and calendar from the Integrations page. Dorinda reads your mail to ground follow-ups in what's actually happened — not guesswork — and reads your real calendar for context. When it's time to send, it stages the message for your approval first; it never auto-sends. Connected account data is used only to provide the feature, never sold, never used for ads or to train a model.

So every follow-up is grounded in what actually happened — not a guess.

  • Read your Gmail or Outlook inbox to ground follow-ups in what's actually happened.
  • Send from your own Gmail or Outlook account — stages a message for your approval first, never auto-sends.
  • Read your Google or Outlook Calendar events and add new ones — your AI can book things for you.
  • Forward (or CC) any email to your private Dorinda address and it lands as an inbox item — a quick capture channel that works from any email client, no app required.
  • Least-privilege and revocable anytime.
Timeline & audit

A record you can trust — everything, and why.

Every delegation carries a plain-language timeline of what happened and why: what the AI read, what it staged, what you approved, and when the loop closed. It is a full audit trail, so accountability is not a promise — it is a record.

So you can always see what happened, and why — no black box, no taking it on faith.

  • Plain-language events, in order.
  • Cited evidence for what the AI grounded on.
  • Your approvals recorded alongside the action.
Household · the expansion

Bring the whole house into the loop, when you're ready.

Start with just you — bring the household in whenever you're ready. On Family, up to five members share a Household — a shared calendar, lists, and projects — with roles for Owner, Adult, Teen, and Assistant. Every item is scoped shared or private, so the family sees what is shared and each person keeps their own private Dorinda.

So the whole house stays on the same page, without any one person holding it all in their head.

  • Up to 5 members, each with the full Personal experience.
  • Roles: Owner · Adult · Teen · Assistant.
  • Shared vs. private scoping on every item.
Everything else in the daybook

The whole operating picture, in one place

Inbox & capture

A single place for everything that comes in. Capture from your AI chat, or forward (or CC) anything to your private Dorinda address — it files as an inbox item and open loop, ready to track.

Calendar & People

Your schedule and the people it involves, in context. Who you owe, who owes you, and what is coming up next.

Projects

Group related work into a project so a bigger effort stays coherent — and you can see it move toward done.

Reminders & routines

Persistent reminders that keep going until the loop is actually closed, plus routines for the things that recur.

Notifications

Nudges and approval requests surface via browser push, on your phone, and by text message (SMS rolling out) — opt-in, quiet-hours aware, so the right things reach you at the right time.

Search

Find anything you’re tracking — a decision, a person, or a note — across your whole record, the memory your AI reads from.

One memory across your AI

Start in ChatGPT, pick it up in Claude. It is the same assistant, the same durable history, either place.

See it work with your own AI.

Connect Claude or ChatGPT in under a minute. 14-day free trial, no card needed.