From the founder

I built Sorenda so my AI would remember what I know

Every chat started from zero. Everything I knew was already written down, just where no assistant could read it.

My name is Onada Nelson. Below are four situations from an ordinary week: how I worked before, and how I work now that my notes are my assistant’s memory. Every card comes from my own notebook.

Start for free Read the stories

Note January 12, 2025

Explained my product to Claude for the fortieth time today. I have 2,300 notes about it. It has read none of them.

Why Sorenda

#idea

Memory that answers

On Fridays I ask Claude what I promised during the week

In meetings I write down who I promised what, and by Friday half of it is buried. Now Sorenda is Claude’s memory: it searches my notebook, reads my tasks and answers with links to my own notes.

It may add new notes, but it only proposes changes to existing ones, and it never sees the notes I hid from AI.

BeforeRetelling the week to a chatbot from scratch

NowOne question, answered from my notes

What did I promise to do after this week’s meetings?

Searched Sorenda: 5 notes, 3 tasks

Send Mark the updated deck from Tuesday’s meeting, due Friday
Rewrite the first-screen copy from the interview with Anya: “not clear where to start”
Invite Anya to the beta done on Wednesday

Search by meaning

I remembered what the conversation was about, but not a single word

In the spring I had coffee with Tom, who has built a team from scratch twice. He said one thing about the first hire, and I wrote it down. Six months later, when I actually started hiring, I couldn’t find it: the note never says “advice” or “hiring”.

Sorenda’s search compares meaning, not letters. I describe what I remember in my own words and the note turns up. My assistant searches the same way.

BeforeGuessing phrasings and scrolling the feed

NowDescribe what I remember

advice on who to hire first ✦ AI

Coffee with Tom …Bring in the person who will do what you keep putting off. You don’t need a copy of yourself. April 14, matched by meaning: “first employee”, “who to bring on”

Job post: developer Draft of the role description, version 2

Four Thousand Weeks …saying no to good things for the sake of the main one

Everything arrives by itself

I stopped retyping things into my notes

Half of what I need to remember never starts as a note: a booking in my inbox, a call with a client, a receipt on the table. I used to copy the important bits by hand, and mostly I didn’t.

Now I forward the email and get a card. The Mac companion records the call and writes the summary with the tasks. A photo of the receipt is text in search. Claude sees all of it next to my own notes.

BeforeRetyping, or losing it

NowForward, record, photograph

1Forwarded the hotel confirmation

Fwd: Your booking is confirmed

Booking

Hotel Baltazar, Lisbon €312

Check-in Thursday

Two nights, breakfast included

2Recorded the call on Mac

Recorded on Mac 24:40

Call with Anya

Anya opens her notes only to search. She will test the digest in the beta.

Invite Anya to the beta
Send her the digest example

AI proposes, I decide

Most of my notes died the day after I wrote them

I would write down an idea and never come back to it. Not because the ideas were bad: nothing connected a new note to the old ones, and I don’t have time to reread everything.

Now an agent reads new notes, finds related ones among the old and proposes links and summaries once a week. Nothing changes until I accept it in my inbox. Claude lives by the same rule.

BeforeIdeas nobody rereads

NowA weekly report and proposals I accept or reject

Weekly report, September 8 to 14

34new notes

6suggested links

3forgotten topics