Context portability needs unusual partnerships
Partnering with Google, Meta, DTI, and regulators so your context can travel and the internet remembers you
The internet knows you in pieces.
Google knows the late-night researcher with 37 tabs open. Instagram sees the social butterfly with the impeccable food pics and Pucci scarves. ChatGPT meets you in work mode.
None of them know you.
Now picture waking up with a $0 personal staff:
A shopper who actually understands your taste. They’ve followed your stories for years and can spot the constants and the change
A concierge who suggests a quiet break in a picturesque Italian village in late September because you hate crowds and temperatures over 30°C
A nutritionist who doesn’t nag for a food diary because they already see the patterns in what you eat and what you post
LLMs have made intelligence cheap. The scarce thing now is memory: your personal context. Intelligence without personal context is just a clever stranger.
So how do you enjoy all these AI services without re-training them in every app, every time?
What if your apps could just…know?
Imagine never rewriting the same elaborate prompts again. No more taking and uploading screenshots. No more rebuilding your preferences from scratch on every AI product. Your context travels with you, on your terms. So every agent, from shopping to travel to health, starts on page 500 instead of page 1.
Here is the tension: why would Meta share your context with Gemini, Gemini share with ChatGPT, or ChatGPT share with Claude? The better your experience becomes inside each product, the less incentive any of them has to let your memory roam.
Intelligence is being commoditised. Memory is becoming the moat. And trying to build 1:1 partnerships between every company that has a fragment of you and every company that needs it is not practical.
The missing piece: a user-owned context layer
There is a role nobody is really playing yet: a user-owned context layer.
This layer would:
Partner with context holders like Google, Meta, and OpenAI to turn scattered traces into portable, structured context
Give you simple controls so you can decide what is shared, with whom, for how long, and revoke access when you want
Provide developers with tooling (APIs, MCPs, agents) so they can build context-rich experiences without brittle one-off integrations
In practice, you set your rules. Apps request access. You can see what moved, when and why, and you can switch it off in a click.
That is what we are building at Fabric.
Read our Manifesto to learn more about our philosophy.
TL;DR: your digital self should be something you carry, not something platforms hoard.
Why this needs unusual partnerships
Thousands of one-to-one context deals are unwieldy for users and impractical for businesses. No one wants to consent repeatedly to sharing context between multiple different apps, repeating the flow for every single context holder and every single context user.
What we need is a small set of high-leverage partnerships that unlock portability, plus sensible regulations that make portability work in practice.
Policy is already pointing the way. In the EU, the Digital Markets Act’s Article 6(9) pushes for continuous, real-time portability tools for end users and the third parties they authorise. In the UK, DSIT is shaping a Smart Data scheme for digital markets, and the CMA is exploring portability for search context.
At the same time, the market is becoming agentic. Google just shared its Conversational Commerce Agent for online stores. Do you want these shopping agents to have amnesia, or to actually understand you based on your activity across the internet?
Portability and partnerships are what make the second option possible.
Our approach to partnerships
Our job is to make portability usable, not just legal. That means pairing product work with real partnerships across platforms, industry bodies, and regulators.
Google
We have worked with Google since the first version of their Data Portability API.
We’ve given hands-on feedback and seen concrete improvements in:
Data quality: clean, structured, well-timestamped records with rich metadata
Latency: reduced from hours to minutes, which expands what you can build
We continue to collaborate with Google to open up other sources of data (for example, Gemini) and to improve the consumer consent experience.
Google also asked Massimo (Co-founder and CTO, Fabric) to present our developer experience with the Data Portability API at the EU Commission. This is not a box-ticking exercise. It is about making portability work for builders and users.
Meta
We have been working with Meta for months and are now listed as a transfer destination in their apps.
In that time, we have seen substantial progress in the API:
Unified flows: bringing “Download Your Information” (DYI) and “Transfer Your Information” (TYI) into a single “Export Your Information” (EYI) experience for cleaner user journeys
Meta AI data availability: Meta AI data is available through the Data Portability API
In addition, Meta’s API works globally, signalling a commitment to worldwide portability.
We are continuing to work with Meta on improving data quality and on further consent UX improvements.
Meta also presented an early version of our product at the EU Commission:
DTI
The Data Transfer Initiative is an industry portability effort supported by partners like Meta and Google.
Portability runs on trust, and DTI is leading the way on an industry-wide joint trust framework. We are certified Trust Level 2 in DTI’s Joint Trust Registry.
Our engineers contribute to DTI’s codebase, and we appreciate DTI nudging AI companies toward real interoperability. Shout-out to Anthropic for engaging publicly on this.
Regulators (EU & UK)
Context portability has already become a reality in Europe through the EU Commission’s Digital Markets Act, Article 6(9).
In the UK, DSIT’s Digital Markets Smart Data scheme and the CMA’s work on search data portability are the levers that will make context portability work for people in the UK.
We have had multiple productive working sessions with these teams and we support their mission to put portability in people’s hands.
Regulatory certainty is essential. Recent uncertainty in US open banking, including policy shifts and proposed fees for access to financial data, saw Visa pull back from US open banking. Outcomes like that reduce consumer choice and set back the ecosystem. The EU and the UK are leading the way by setting clear rules from the start.
The personalized Internet is (almost) here
A personalized internet is simple in idea and hard in practice. But it’s within reach. It needs strong partnerships, clear policy, and context-rich products people actually want to use.


This really resonated — especially the line “you set your rules, apps request access.”
That feels like the quiet hinge in the whole piece.
Once apps are requesting access in real time, the problem stops being just context portability and starts looking like an execution question: when and how permission is checked before something actually happens. In an agentic world, memory moving isn’t enough — actions need a place to ask first.
What’s interesting is how naturally your partnerships, trust registries, and regulator work point in that direction. Making portability usable (not just legal) seems to require a runtime layer where intent, scope, and consent are enforced at the moment of execution — not inferred after the fact.
This feels less like a departure from what you’re building and more like the next inevitable layer above it. Really strong piece.