Every system that learns something about you from your behaviour is accumulating something intimate. What you read, what you pause on, what you return to, what you avoid. The shape of your attention over time. The pattern of what you need and when you need it. The record of who you have been across the months and years you have spent inside it.

That knowledge is among the most personal a system can hold. And almost universally, it is held by the company that built the system, managed in ways the person it concerns cannot see, and used toward ends that serve the platform before they serve the person. The data that accumulates about you is not yours. It is theirs. That is not an accident of implementation. It is a structural choice, made deliberately, because the business model depends on it.

Memory is where this condition is most acute. A system with genuine episodic memory, one that can reconstruct not just what you said but the structure of your experience over time and the arc of your concerns as a person, has access to something more intimate than most institutions are ever trusted with. The question of who holds that, and what they are structurally permitted to do with it, is not a privacy question in the ordinary sense. It is a question about what a person is and what any system that touches their continuity owes them.

Mpalo exists because the answer currently being given to that question is wrong. Not wrong as a policy failure that better regulation could fix. Wrong about something more fundamental. Wrong about what user data is, who it belongs to, and what the architecture of a system that touches it should refuse to allow regardless of who asks.

The founding belief is this: intimate technology should not be held or managed by companies on behalf of the people it concerns. It should be made accessible, so that the person it belongs to can hold it themselves. Memory that concerns a person should be held by that person. Everything Mpalo builds is an attempt to make that structurally true rather than aspirationally stated.

The founding belief at Mpalo is not a product feature and it is not a policy. It is a structural commitment that shapes every architectural decision the company makes, expressed in governance that does not bend under commercial pressure and in code that makes certain outcomes impossible rather than merely unlikely.

Episodic memory is intimate in a way that ordinary data is not. A system that can reconstruct the structure of your experience over time holds something closer to your continuity as a person than to your browsing history. We believe that the only honest response to that intimacy is an architecture that refuses exploitation by design. Not by promise. By design.

This is why Bring Your Own Vector Store is in the core of Palo Bloom rather than an enterprise afterthought. It is why the Open Decay Protocol is being formalized in our legal structure before launch rather than described in a blog post after the fact. The protocol is this: if Mpalo is ever acquired, all user memory data is automatically deleted. It does not transfer. It is why the non-weaponisation commitment is charter-level, protected by Class M shares and a Foundation observer seat, rather than a terms of service clause subject to quiet revision. The restraint is in the structure. Not the policy.

Mpalo is incorporated as a Delaware Public Benefit Corporation. The PBC structure creates a legal obligation to balance the interests of shareholders against the stated public benefit purpose and the interests of those materially affected by the company's conduct.

• Public Benefit Purpose. Our stated purpose is the development of memory infrastructure that is consent-driven, honest about what it is and is not, and structurally resistant to weaponization.
• The Open Decay Protocol. If Mpalo is ever acquired, all user memory data is automatically deleted. It is not transferred. You cannot acquire Mpalo and receive its users' memories. This will be formalized in our legal structure before launch.
• Non-weaponization. Mpalo will not develop or deploy cognitive manipulation systems targeting civilian populations. This is a charter-level constraint, not a policy subject to revision under commercial pressure.
• Public Benefit Report. We will publish an assessment of our performance against stated public benefit objectives at minimum every two years. The report is the primary external verification instrument for our PBC obligations.