Your data stays yours. Whether you are building as an individual developer or deploying
at scale, your information remains private within your chosen environment. Palo
processes your data to build memories. Your raw content is never permanently retained
by us. You can view, export, and delete your memories at any time.
Making AI Remember
Most AI systems treat memory as a retrieval problem. They extract facts from conversations,
store them, and retrieve the most similar ones when you ask. This is useful. It is not memory.
Human memory is not a database of facts. It is a continuous thread of experience: events,
contexts, emotions, and personal meaning woven together over time. Palo is built on this
understanding. It encodes experience rather than extracting facts, and it can traverse a user's
history rather than search it. The difference is not a feature. It is the reason Mpalo exists.
Built for Builders and for People
Palo is infrastructure for developers building agents and copilots that need genuine memory, not
fact retrieval dressed as personalisation. It is also, eventually, for the people those agents
serve: anyone whose relationship with an AI system depends on that system actually knowing them
over time.
We offer flexible API pricing for developers and growing teams. Custom deployments for larger
organisations. The goal is that the quality of memory a product offers its users shouldn't depend on the size of the team building it.
Your Privacy
We do not train our shared base models on your private data without your explicit consent. Palo
uses a personal memory layer that adapts to you within your own memory space. Your memories do
not modify shared model weights.
All data is encrypted in transit and at rest. You own what you put in. You can delete it
permanently at any time.
Mpalo is incorporated as a Delaware Public Benefit Corporation. The mission is not a marketing
position. It is a governance structure. The board is required to balance commercial returns
against our stated public benefit: developing memory and cognition technologies transparently,
deploying them only with consent, and refusing to allow them to become instruments of
psychological manipulation.
We decline partnerships that require compromising those constraints. We reinvest a significant
portion of profits into research on memory, cognition, and ethical AI infrastructure. We will
not weaponise what we build, regardless of who asks.
Our Visual Philosophy
Our visual philosophy is a direct reflection of what Palo does. We draw from natural forms:
flowers and fishes (yes fishes, not fish), organic, living, imperfect. Memory is not a filing
system. Our icons are often abstract, barely resolved, more impression than illustration. That
ambiguity is intentional. It mirrors how memory actually works: retaining the significant
essence of a moment rather than a perfect record of it.
Our assets move. Not for decoration, because memory is not static. Memories are active
reconstructions, slightly different with every recall. The constant, gentle motion in our design
captures the fluidity of thought and the imperfection of memory we are working to understand and
embody in Palo.
What We Are Building Toward
Palo Bloom is in development. PaloSpring, our open-source agentic coding runtime, is coming
soon. The research programme, three papers published, more in progress, is the public record of
the thinking underneath the product.
We are building toward a future where AI genuinely understands the people it works with. Not
through surveillance or optimisation, but through the kind of memory that makes sustained
understanding possible. That is a harder problem than fact retrieval. It is the one worth
solving.
