I’m launching Context Passport: a portable AI context layer you control
If you use more than one AI tool, you have probably experienced a strange kind of repetition.
You explain your preferences to ChatGPT. Then you open Claude and explain them again. You use Gemini for another task and start over. Perplexity may be excellent for a particular research question, but it does not automatically inherit the context you already established somewhere else.
AI assistants are getting better at personalization. ChatGPT Memory can use useful context from past chats, files and connected apps when memory is enabled. Gemini can also personalize responses using memory of past chats.
Those capabilities are useful. But they raise a bigger question:
Should the durable context that helps AI understand you live inside one AI provider, or should it belong to you?
That is the problem behind Context Passport, which I’m launching as a separate, user-controlled context layer for the AI tools people already use.
The short answer: what is Context Passport?
Context Passport is a personal AI context management service that lets you keep preferences, goals, constraints and memories in one place, then selectively share relevant context with supported AI tools.
At launch, Context Passport supports workflows with ChatGPT, Claude, Gemini and Perplexity. Manual sharing is the default. You see what is about to be shared and can remove anything before it leaves your Context Passport profile.
The idea is simple:
Teach once. Choose what travels.
That is why I describe the product as portable AI context, rather than simply another memory feature.
Why AI context is becoming a portability problem
The usefulness of AI often depends on context.
Imagine asking for a laptop recommendation. A generic answer may be technically correct, but it becomes more useful if the AI already knows:
- your maximum budget is €2,000;
- battery life matters more than raw performance;
- you mainly work with marketing and AI tools;
- you already use Apple devices.
The same applies to professional work.
An AI response can become more relevant when it knows your role, preferred level of detail, recurring goals, project constraints, audience, decision criteria or formatting preferences.
The friction appears when that information is fragmented.
One assistant knows how you like answers structured. Another knows the project you are working on. A third has useful context from a previous conversation, but only inside that platform.
Some AI providers are starting to address this. Google, for example, now documents a feature that lets eligible users import memory and chat history from other AI platforms into Gemini. That is useful, but an import is still different from maintaining one independent context source that you control and can selectively use across several AI products.
Context Passport is built around that distinction.
Why I care about this problem
My main work is still search growth for SaaS and B2B companies. I work directly across Google Ads, commercial SEO and AI Visibility, with a particular focus on how people and systems find, understand and evaluate information.
That same theme appears in my AI Visibility work: useful information needs to be accessible, explicit, well-structured and attached to the right entity and evidence.
Context Passport applies a related principle at the individual level.
The problem is no longer only, “Can an AI system find the information?”
It is also:
“Can I decide what context an AI system receives about me, and can I carry that context between tools without rebuilding it every time?”
Context Passport is my attempt to make that easier.
What does “portable AI context” mean?
Portable AI context is user-controlled information that can be reused across different AI tools instead of remaining trapped inside a single assistant or conversation history.
That context can include:
- preferences;
- goals;
- constraints;
- profile information;
- confirmed memories;
- recurring working rules;
- other information you deliberately decide is useful across AI interactions.
This is broader than memory alone.
A memory is usually something an AI system retained or inferred from previous interactions. Context can also be something you define intentionally before any model has inferred it.
For example:
Keep software recommendations below €100 per month unless I explicitly say otherwise.
That is useful context even if no AI assistant “learned” it from a previous chat.
How Context Passport works
The product is designed around three steps.
1. Keep useful context in one private profile
You create a profile containing the information you regularly find yourself explaining to AI tools.
You do not need to recreate your entire digital life.
The useful starting point is the context that repeatedly changes the quality of an answer.
Examples might include:
I work primarily with B2B SaaS companies.
I prefer concise answers with concrete examples.
This project targets enterprise buyers, not small businesses.
Battery life matters more to me than maximum benchmark performance.
Avoid recommendations above this budget unless there is a clear reason.
The goal is not maximum context.
The goal is minimum useful context.
2. Context Passport selects what is relevant and permitted
Not every stored fact belongs in every prompt.
Your dietary preferences probably do not matter when you are debugging code. A professional project constraint probably does not belong in a personal travel question.
Context Passport first applies permission and category boundaries, then considers relevance.
Its security documentation describes this as a minimum-disclosure approach: permission scopes and category restrictions are checked before relevance ranking, and missing permission removes an item from consideration rather than simply giving it a lower score.
That matters because more context is not automatically better context.
Useful personalization depends on the right context being present, not on every possible fact being exposed.
3. You preview what will be shared
Before approved context is added to a prompt, Context Passport shows a sharing preview.
You can uncheck anything you do not want to include.
According to the Context Passport privacy policy, only the context items you leave checked are added to the prompt on the supported AI website.
In the Chrome extension workflow, Context Passport reads the current prompt only after you click Personalize. The extension does not continuously upload conversations, and it does not click Submit for you.
That human review step is intentional.
Context Passport is not another AI assistant
Context Passport does not replace ChatGPT, Claude, Gemini or Perplexity.
Those products remain the places where you ask questions, generate content, research, reason or work with models.
Context Passport sits at a different layer.
It is the place where you keep the approved personal context you want to make reusable.
A useful mental model is:
The AI provider owns the conversation. You own the context.
That is also why Context Passport does not include access to paid AI subscriptions. Its pricing page explicitly states that it carries approved context into the AI services you already use rather than replacing those services.
Why Context Passport uses selective sharing instead of invisible collection
Personalization becomes uncomfortable when you do not know what information is being used, where it came from or why it appeared in a response.
I wanted Context Passport to make those boundaries visible.
The product lets users inspect and manage profile data and memories, including actions to edit, archive, reject, export or delete information.
The privacy policy also states that Context Passport does not, by default:
- collect browsing history;
- continuously upload AI conversations;
- sell personal profile data;
- use profile data for advertising;
- use profile data to train unrelated models.
The browser extension stores scoped authentication information on the device and uses the current prompt only after the user actively requests personalization.
There is also an optional AI-assisted ranking capability on paid plans. When a user enables it, selected query and candidate context may be processed server-side by OpenAI for ranking. I think that limitation is worth stating clearly rather than hiding it behind a vague “privacy-first” claim.
Specific claims are more useful than slogans.
What security controls does Context Passport use?
No software service can truthfully promise absolute security, and Context Passport does not make that claim.
Its security page documents several implemented controls, including:
- Supabase Auth for web sessions;
- OAuth 2.1 authorization code flow with PKCE for remote MCP clients;
- scoped, short-lived extension tokens;
- Row Level Security on user-owned database tables;
- server-only service credentials;
- permission checks before relevance ranking;
- environment-based secret management;
- request limits, schema validation and structured logging.
The point of listing these controls is not to claim that risk disappears.
It is to make the product’s boundaries inspectable.
Cross-AI memory and portable AI context are related, but they are not identical
You will increasingly see phrases such as:
- cross-AI memory;
- shared AI memory;
- portable AI memory;
- AI memory portability;
- personal AI memory;
- AI context layer.
These phrases point toward a similar problem: people are using multiple AI products, but useful personal context is often fragmented across them.
Context Passport is intentionally centered around user-controlled context records and selective disclosure.
That is why I currently prefer portable AI context as the broader description.
Memory is part of the product, but context can also include explicit information you chose to define yourself.
A budget is context.
A communication preference is context.
A role is context.
A current goal is context.
A project constraint is context.
You should not necessarily have to wait for an AI assistant to infer those things before they become reusable.
How is Context Passport different from built-in AI memory?
Built-in AI memory is useful.
OpenAI’s current Memory documentation explains how ChatGPT can remember useful context to personalize future responses. Google’s Gemini documentation similarly describes personalization based on memory of past chats.
Those features solve personalization inside their own product experiences.
Context Passport addresses a different layer:
What if the stable source of important personal context sits with the user, while individual AI tools become destinations for selected context?
This is not an argument that native AI memory is bad.
It is an argument that native memory and portable context solve different problems.
Native memory can make one AI assistant better at knowing you.
Portable context can help you decide what knowledge should travel when you move between assistants.
Who is Context Passport for?
I think the product becomes most useful when you regularly move between AI tools.
Knowledge workers using several AI assistants
You may prefer one model for research, another for writing and another for analysis.
Your basic preferences and recurring constraints should not need to be rewritten every time.
Consultants and marketers
Audience constraints, positioning decisions, formatting preferences and project rules can affect the usefulness of an AI response.
For people doing client work, a reusable but permissioned context layer can reduce repetitive briefing.
Founders and operators
Business priorities, product constraints, current goals and decision rules often remain relevant across many AI tasks.
Developers and technical users
The Pro plan includes permissioned MCP access, allowing authorized MCP clients to access context through scoped permissions.
People who want personalization with more explicit control
Some users want AI personalization but are uncomfortable with context being gathered or reused invisibly.
Context Passport is designed around preview, permissions, editing, export and deletion rather than automatic disclosure of everything stored.
Is Context Passport free?
Yes.
The current Context Passport pricing includes a free plan with no payment details required.
At launch, the free plan includes:
- 30 personalizations each month;
- up to 25 saved preferences, goals, constraints and memories;
- ChatGPT, Claude, Gemini and Perplexity support;
- preview and approval before sharing;
- export, deletion and privacy controls.
The current Pro monthly plan is €4.99 per month, and the annual plan is €39 per year.
Paid plans add features including:
- unlimited personalizations under fair use;
- unlimited saved context records;
- AI-assisted memory suggestions;
- permissioned MCP access;
- optional AI-assisted ranking;
- up to 730 days of activity history.
Because pricing and product limits can change, the live pricing page should remain the source of truth.
What Context Passport does not do
Clear product boundaries are important, especially for a tool handling personal context.
Context Passport does not:
- guarantee that an AI response will be correct;
- control how a third-party AI provider processes text after you send it;
- replace ChatGPT, Claude, Gemini or Perplexity;
- guarantee absolute security;
- continuously collect your AI conversations by default.
Those limitations are not weaknesses to hide.
They are part of understanding what the product actually does.
Why launch this now?
AI products are moving steadily toward deeper personalization.
ChatGPT now has a more developed memory system designed to reduce repetition across interactions. Gemini can personalize from past chats, and Google now offers eligible users a way to import memory and chat history from another AI platform.
That direction makes the underlying ownership question more important, not less:
Where should the durable version of your AI context live?
Inside whichever assistant you use most today?
Copied manually between tools?
Imported from one provider into another?
Or maintained in a separate layer that you control and selectively share?
Context Passport is built around the fourth option.
Context should travel with the person, not the platform
The web became more useful when important things became portable.
Contacts became portable.
Calendars became portable.
Documents became portable.
Authentication became more interoperable.
I think personal AI context is beginning to face a similar question.
That does not mean every conversation should travel.
It does not mean every memory should be shared.
And portability should not mean giving every AI tool unrestricted access to everything another tool has ever learned about you.
My view is narrower:
Useful context should be portable when the user wants it to be.
You should be able to see it.
Correct it.
Delete it.
Restrict it.
And choose where it goes.
That is what I want Context Passport to make easier.
If this problem sounds familiar, you can create a free Context Passport profile and try it with the AI tools you already use.
Teach once. Choose what travels.
FAQ: Context Passport, portable AI context and cross-AI memory
What is Context Passport?
Context Passport is a user-controlled personal AI context layer. It stores approved information such as preferences, goals, constraints and memories, then lets users selectively share relevant context with supported AI tools.
What is portable AI context?
Portable AI context is reusable, user-controlled information that can travel between different AI tools instead of remaining available only inside one assistant, account or conversation history.
Which AI tools does Context Passport work with?
At launch, Context Passport lists support for ChatGPT, Claude, Gemini and Perplexity.
Can ChatGPT and Claude share the same memory through Context Passport?
The more precise description is that Context Passport can act as a common source of approved context that you choose to use with supported AI tools. It does not claim to directly synchronize the native memory databases of ChatGPT and Claude.
Does Context Passport automatically collect my AI conversations?
No, not by default. The privacy policy states that Context Passport does not continuously upload AI conversations or collect browsing history. In the supported browser workflow, the current prompt is processed after the user explicitly clicks Personalize.
Can I see what Context Passport shares with an AI?
Yes. Context Passport is designed to show a preview of the selected context before sharing. Users can remove items before approved context is inserted or copied.
Is Context Passport the same as ChatGPT Memory?
No. ChatGPT Memory is part of ChatGPT’s own personalization system. Context Passport is a separate context layer intended to make approved information reusable with multiple supported AI tools.
Does Context Passport replace my ChatGPT, Claude, Gemini or Perplexity subscription?
No. Context Passport carries approved context into AI services you already use. It does not include access to those services or replace their subscriptions.
Is Context Passport free?
Yes. The free plan currently includes 30 personalizations per month, up to 25 saved context items, support for ChatGPT, Claude, Gemini and Perplexity, and the core preview, export, deletion and privacy controls.
Who is launching Context Passport?
Context Passport is being launched by Andrei Visan, a search growth consultant working across Google Ads, commercial SEO and AI Visibility for SaaS and B2B.
His broader SEO + AI Visibility work focuses on making useful information easier for buyers, search engines and AI systems to find, understand and evaluate.