Writing and research
Research that remembers what you already read
The waste in research is not reading. It is re-finding what you already read.
- Situation
- Working on something long, over months, in the gaps between other work.
- Brought in
- Sources, papers and saved articles
- Your own notes and half-formed thinking
- History imported from other AI tools you have used along the way
- First ask
- “Pull together everything I have gathered on this, and tell me what I have not resolved.”
- Result
- A view of your own material that you could not hold in your head: what you have, what it says, and where the argument still has a hole in it.
- Why it helps
- The thinking accumulates in one place rather than across a folder, three apps and two abandoned assistants.
How it unfolds
Bring the scattered material together
Sources, notes and past AI conversations, including exports from tools you have moved on from.
Ask what you already know
Before reading anything new. It is usually more than you remember.
Delegate the sifting
Set a research or review task going and come back to something readable.
Talk it through when it is still messy
Voice suits the stage where you do not yet know what you think, and want to find out by saying it.
Further questions in this journey
- “What have I already written about this?”
- “Where does my own material contradict itself?”
- “What am I avoiding in this argument?”
Capabilities that carry this journey
Switching from another AI
Bring your history from other AI tools so your assistant does not begin at nothing.
Memory, notes and glossary
Notes you deliberately capture, plus a glossary that keeps your world consistent.
Delegating work
Research, drafts and analysis become tasks with a lifecycle, not one-off replies.
Voice
Natural spoken conversation, for hands-free moments and thinking out loud.
Background reading: what a personal ai should actually remember.