The most useful thing you can give a capable AI isn’t another chatbot to learn — it’s follow-through: the spine to stay on what you hand it, the memory to carry it across chats, and the rules it follows. Dorinda plugs into the ChatGPT or Claude you already use and makes it an assistant that actually follows through.
Today’s AI is genuinely sharp in the moment. Ask Claude or ChatGPT to think through a problem and it will. Then the thread scrolls away, the memory doesn’t carry to the next chat, and the follow-through lands back on you.
The intelligence is real; the accountability is missing.
Dorinda is a complete, deterministic system of record that the ChatGPT or Claude you already use plugs into over MCP — no new chatbot to learn, and no lock-in. It gives your AI three things it can’t give itself: real accountability, persistent memory, and the controls you set. Your AI does the thinking; Dorinda gives it the spine to follow through and the memory to make it stick.
We built everything around the Delegation — a durable, closed-loop responsibility that is tracked until it’s actually done, with who owes what, what’s waiting, and what’s at risk. A to-do app remembers a line of text. Dorinda closes the loop, and keeps a plain-language timeline of everything that happened and why.
Delegating to software only works if you can trust it not to overstep. So the rules are yours: sending an external message, spending money, or contacting someone new always waits for your approval — right in the chat — while the low-risk stuff can be auto-allowed to stay out of your way. Your assistant state belongs to you, across every AI you use, and every action is on the record.
We’re building Dorinda with and for people whose brains don’t do “out of sight, out of mind” gently — starting with the ADHD and executive-function community. We don’t claim to have all the answers; we listen hard and build what actually closes the loop.
Start with just you. Dorinda is built for individuals first — and, whenever you’re ready, for households too: a shared place where a family can keep the loop closed together, while everyone keeps their own private Dorinda. The goal is simple: make the AI you already trust actually follow through.