Product / Memory
It remembers, and you can read exactly what.
Durable facts about you and your work carry into every conversation, applied quietly instead of re-explained. You can open Memory any time and see the full list.
Memory
You can edit or delete any of this
Ships on Thursdays; avoid scheduling reviews after 3pm.
Working styleEditDeletePrefers written summaries over meetings for status.
PreferenceEditDeleteNorthwind is the largest account; Dana is the main contact.
ContextEditDelete
Praxa keeps durable facts about you and your work and applies them quietly, without you re-explaining yourself. Open Memory whenever you like to see the full list, grouped into people, projects, routines, open loops, and decisions, and add, edit, or delete anything in it.
What it does
A graph that maintains itself
One graph, not a pile of notes
People, projects, routines, and decisions connect into a knowledge graph that gets smarter as they interact.
Cleans up after itself
Contradictions get superseded, duplicate people merge, and time-bound facts expire on schedule, without you doing the tidying.
One switch, total control
Turning memory off stops new writes and old reads at once, and deletion removes the derived model, not just the raw entry.
How it works
Step 1: It notices, quietly
Durable facts get captured from ordinary conversation, not a separate interview.
Step 2: It organizes itself
People, projects, and routines link into a graph; outdated facts get superseded and time-bound ones expire.
Step 3: You stay in control
Open Memory to read, edit, or delete anything, or ask "what do you know about me" for the real answer.
What it does not do
- Memory is one switch away from off, and turning it off stops both new writes and old reads.
- Health and camera data never enter memory.
- Passwords, API keys, and other secrets are refused at the door; they cannot be stored.
- This is memory of you and your work, not a company wiki or a document search engine.
Frequently asked questions
What is organizational memory and why does it matter for AI?
It is durable knowledge that carries forward instead of resetting every conversation. Praxa builds it from ordinary conversation, organizes it into a knowledge graph of people, projects, and routines, and labels inferred facts so you always know why something is there.
How do AI agents remember things about me and my work?
Praxa captures at most a couple of durable facts per qualifying conversation, with no added wait time, and reviews what it captured afterward to keep it accurate. The same pre-rendered memory feeds chat, the daily greeting, and automations consistently.
Can I see everything an AI remembers about me?
Yes. Open the Memory screen to see the full list, grouped into people, projects, routines, open loops, and decisions, and add, edit, or delete any entry yourself. You can also just ask "what do you know about me" in chat and get the real answer, not a canned reply.
Is there an AI memory system I can turn off completely?
Yes. One switch turns memory off for real: it stops new facts from being written and stops Praxa from reading what it already stored, not just from adding more. It is the one Capabilities toggle that persists and genuinely changes behavior.
Does AI memory ever store passwords or sensitive data?
No. Passwords, API keys, and similar secrets are refused before they can enter memory at all, and health or camera data never enters it either. What gets stored is durable facts from conversation, labeled by where they came from.
See what it already knows.
Download for iPhone, or open Praxa on the web. Free while we are in beta.