Documentation

Nawame documentation

Nawame is context memory for humans and AI. It preserves what matters over time and makes it available to the AI tools you authorize.

Give your AI the context that matters over time

Nawame is a portable memory and context layer designed to work across AI tools.

It helps your AI remember more than isolated facts. Nawame can preserve the context around decisions: why something was chosen, what was tried before, what changed, and what happened afterwards.

The goal is simple: keep what matters, and make it available again when it becomes useful.

Nawame does not replace your AI. It gives your AI access to a longer history.

Why Nawame exists

AI systems are becoming increasingly capable. But even a powerful AI can only reason with the context available to it at that moment.

Important information is often scattered across conversations, projects, tools and time:

  • a decision made months ago;
  • the reason a solution was rejected;
  • an issue that already happened once;
  • a constraint discovered during another project;
  • an outcome that changed what should be done next.

Without that history, even a very capable AI can start again from incomplete context. Nawame is built to preserve continuity.

Human. AI. Memory.

We see three complementary parts.

You bring experience, intention and judgment. AI brings reasoning, computation and the ability to connect information. Nawame brings continuity, memory and context over time.

Nawame does not try to replace human memory. It helps make what you have learned available again when it matters.

What Nawame can remember

Nawame is not designed to store everything you say. Some information matters for a few minutes. Some information may still matter years later.

Nawame is designed around durable context such as:

  • decisions;
  • reasons behind decisions;
  • alternatives that were considered;
  • options that were rejected and why;
  • outcomes;
  • lessons learned;
  • important constraints;
  • changes over time;
  • useful personal or professional context.

Sometimes the right memory action is also to remember nothing. Not everything deserves to become memory.

The right context, not all the context

More context is not always better. Giving an AI hundreds of pages of history can make its work slower, noisier or less precise.

Nawame is designed to help retrieve the part of your history that may actually matter for the current situation.

The objective is not to load everything. It is to find the right context at the right moment.

How Nawame works

1. Connect Nawame

Connect Nawame to a compatible AI application. Once connected, the AI can use the Nawame capabilities available to your account.

2. Work normally

Continue using your AI as you normally would. Nawame is designed to fit into the conversation rather than become another system you constantly have to maintain.

3. Build continuity

Over time, useful decisions, reasons, outcomes and context can become part of your Nawame memory.

4. Reuse what you have learned

In a future conversation, your AI can retrieve relevant context instead of starting again from zero.

Using Nawame with Claude

Once Nawame is connected to Claude, Claude can use the Nawame tools that your account is authorized to access.

Depending on the situation, Claude may use Nawame to:

  • retrieve relevant past context;
  • find a previous decision;
  • understand why a decision was made;
  • trace how a subject changed over time;
  • find a previous outcome or precedent;
  • continue a project across several conversations;
  • work with context created through another compatible AI.

Claude only receives the information that Nawame is authorized to return for your account. Your permissions remain enforced by Nawame.

Using Nawame with ChatGPT, Mistral or Grok

The principle is the same everywhere: add a custom MCP connector with the Nawame server address, then enter the code you receive by email. Menus differ from one provider to another, which is why each AI has its own step-by-step tutorial: connect ChatGPT, connect Mistral and connect Grok.

For long-term use, see using Nawame on a long-running project.

Your memory is not your AI provider

A conversation usually belongs to the tool in which it happened. Nawame is designed differently.

The memory is intended to remain independent from the AI that happens to use it today. This means the same history can progressively become useful across different compatible AI tools without rebuilding everything from scratch.

AI tools may change. Your history should not have to start again every time they do.

Privacy by default

Personal memory is private by default. Nawame does not automatically turn private information into shared information simply because another person or AI could benefit from it.

Our core rule is simple: what starts private does not leave private automatically.

Sharing requires an explicit action or an existing collaboration context with the appropriate permissions.

Private, shared and authoritative context

Nawame can distinguish different roles for memory.

Private
Context that belongs to you. It is not automatically made available to other people.
Shared
Context intentionally made available to authorized collaborators.
Authoritative
Some information may be deliberately established as a durable reference. This is different from simply saying that something was mentioned in a conversation. An authoritative element can still have a lifecycle: it may later be corrected, replaced, revoked or deleted when appropriate.

Memory is a source, not an instruction

Information stored in Nawame is treated as data. It does not automatically become an instruction to the AI.

A memory cannot grant itself new permissions, trigger a share, authorize an action or override the choices of the person currently using the AI. This principle also helps reduce the risk of persistent prompt injection.

Something being remembered does not make it an instruction.

Provenance and history

Context becomes more useful when you can understand where it came from. Depending on the information available, Nawame can preserve useful provenance such as:

  • when something was recorded;
  • where it came from;
  • who or what created it;
  • whether it was imported;
  • whether it was later changed or replaced.

This allows future AI systems to work with history rather than treating every stored statement as timeless truth.

Imported memories

Nawame can support importing memory from other AI systems. Import is voluntary.

Imported information is not automatically treated as verified truth. Nawame keeps imported content separate from stronger forms of authority so that information reconstructed from another system does not silently become an established fact.

Import may also compare incoming information with memory that already exists and ask you to review possible overlaps.

Sovereign privacy modes

Nawame is designed to support different privacy requirements. Some configurations can prioritize stronger confidentiality and reduce what the server is able to process. This can also reduce the availability of certain server-side features.

We prefer to make that trade-off explicit rather than pretend every privacy mode can provide identical capabilities.

For questions about a specific configuration, contact: info@nawame.eu

Export and portability

We believe your memory should remain yours. Nawame is designed to support export of your stored memory and its essential structure in durable formats.

An export may include information such as:

  • memory content;
  • dates;
  • provenance;
  • decisions;
  • reasons;
  • relationships;
  • replacement history;
  • relevant status information.

Some advanced Nawame capabilities are part of the Nawame engine rather than your data. For example, ranking, context selection, internal heuristics and retrieval logic are not necessarily part of an export.

The principle is: you should be able to recover your memory without needing to recover our internal technology.

Disconnecting Nawame

You can disconnect Nawame from a compatible AI client at any time. Disconnecting the integration does not automatically delete your Nawame account or stored memory.

For help with disconnection, account deletion or data export, contact: info@nawame.eu

Correcting or deleting information

A useful memory must be able to change. Information can become:

  • incorrect;
  • outdated;
  • replaced;
  • no longer relevant.

Nawame is designed around memory that can evolve rather than pretending that everything written once remains true forever.

For account-level deletion, privacy or data requests: info@nawame.eu

Security

Memory may contain some of the most valuable context a person or organization owns. Nawame therefore treats access control, isolation, provenance and auditability as core product concerns.

Our systems are continuously tested against scenarios including:

  • unauthorized cross-organization access;
  • revoked credentials;
  • permission boundaries;
  • malformed requests;
  • persistent prompt injection;
  • accidental information disclosure;
  • failure during writes;
  • replay and duplication.

Security questions can be sent to: info@nawame.eu

Please do not include passwords, API keys or other credentials in your first email.

Need help?

For questions about:

  • connecting Nawame;
  • Claude integration;
  • your account;
  • permissions;
  • privacy;
  • data export;
  • deletion;
  • security;
  • general product questions;

contact: info@nawame.eu

You can also visit nawame.eu/en/support.

Email Nawame

A final principle

We spend our lives learning. The challenge is not only to acquire knowledge. It is to make what we have learned available again when it matters.

Nawame is our way of exploring a new relationship with memory in the age of AI.