PRODUCT · FOUNDATIONS

Trust is architecture, not policy.

Consent, scope, and purpose are enforced where the data moves — not promised in a document.

Begin the conversation

Principal alignment

PadaMaster works for the organization. PadaOne works for the person. The boundary is structural: the two agents hold different scopes, answer to different principals, and no path lets one quietly optimize for the other. Trust fails when a system pretends to be neutral while serving another principal; Padanet does not pretend.

What reaches the organization is what a person releases, aggregated. Leader views are designed to carry no individual data, and the organizational view needs no personal capture at all.

Sovereignty and governance

European by default, aligned with the strictest regime: tenant isolation and governed interfaces, with nothing crossing to another tenant that you have not released; self-hosted, or hosted in a territory you name. Personal signals sit in the person’s own encrypted channel, designed so that neither their manager, nor their IT, nor Padanet reads them. Confidentiality limits are explicit and agreed in the scoping phase.

Every output is explainable — which evidence, which confidence, which purpose — and reviewable by the people it concerns. Purpose is enforced at the moment of access: a claim usable for staffing is not usable for a performance review, and that is compiled in rather than audited afterwards.

Deployment shapes

Today

Where Padanet runs, and who operates what. One shape is deployed; the rest are specified and are not.

  • Padanet-hosted — managed by us, in Europe. This is what runs today, and what a design partner gets.
  • Private cloud, single tenant — your cloud, your control, our software. Designed; the operational responsibility moves to you.
  • On-premise — fully isolated, for regulated estates, usually with a private exchange node. Designed.
  • Hybrid — organizational nodes in your infrastructure, personal memory private, exchange nodes mediating. Designed, and the shape we expect most large estates to want.

Where data sits

Today

Tenant data lives in a private realm and is not pooled. Two things are shared on purpose, and saying so is the point of this block.

  • Your memory and your agents run on per-tenant nodes in a private realm — the placement rule for anything tenant-owned.
  • Each person's own signals sit in their own encrypted channel, designed so that neither their manager, nor their IT, nor Padanet reads them.
  • Shared on purpose #1: an exchange node in the shared realm, which is how a tenant shares what it chooses to share. Nothing reaches it that the tenant has not released.
  • Shared on purpose #2: model inference may run on a shared pool, for cost. It is stateless — no conversation storage, no fine-tuning side effect, nothing a later request depends on — and a tenant that wants inference private can have it.

What crosses a boundary, and what does not

Today

Connectors read in. The context packet goes out under a declared purpose. The raw signal stays where it was collected.

  • Connectors read from the systems you already own, read-only and purpose-bound. Metadata first: no keystroke capture, no screen capture, no message content.
  • What leaves the memory is a context packet — the smallest adequate, authorized, current, evidence-backed set of claims for one declared purpose — not the signals it was distilled from.
  • Purpose is enforced at the moment of access rather than audited afterwards: a claim usable for staffing is not usable for a performance review.
  • Model calls carry the packet, not the corpus, and are stateless by design.

The aggregation model

Designed

How an individual signal becomes an organizational answer without becoming a readout about a person. Specified, and not something you can inspect on a deployment today.

  • Leader-facing views are aggregated, with a floor below which a cohort is not reported at all.
  • Individual data never becomes an organizational readout: the consented lens is the person's, and what a leader sees is the aggregate.
  • Where a self-view is exposed, the person sees it first and can correct it.
  • This is the part of the architecture a security review should press hardest on, and the part we most want proved on a real estate rather than asserted here.
Where Padanet runs, and what crosses betweenYour systems feed a per-tenant private realm holding the memory and its agents. Out of it goes a context packet under a declared purpose — to your agents, and to a shared exchange node for what you choose to share. Raw signals do not leave the private realm.Your systemsTickets · commitsDocs · calendarsread-onlyPRIVATE REALM — PER TENANTThe memoryclaims · evidence · freshnessYour agentsmissions, on your dataEach person’s own signals: their own channelraw signal does not leave this boxThe agents you runAPI · MCPcontext packet onlyShared realmExchange nodeModel pool, statelessonly what you releasesignalspacketpacket
The trust boundary. Signals are read in; what leaves is a context packet under a declared purpose. The shared realm is dashed because a tenant reaches it only with what it has chosen to release.
What is marked Designed above is specified and not yet deployed: today Padanet runs as the hosted shape, and proving any other one on your estate is part of the partnership. The privacy policy holds the commitments currently in force.

For your agents

Context for agentic AI splits into three layers: data context (can the agent find the right information), action context (can it act safely), and organizational context — how this company actually works. Padanet is the third. Your agents already need to know it, and Padanet gives them a governed way to ask. An agent requests context for a declared purpose and receives a context packet: the smallest adequate, authorized, current, evidence-backed set of claims for that task. Purpose-bound access is compiled in — permissions shape what is assembled, rather than filtering a result that already crossed a boundary.

  • Relevant claims, with the evidence behind them
  • Policy notices — what each element may be used for, travelling with the packet
  • Confidence, contradictions and omissions stated as such
  • Freshness — how recent the evidence is, and where it has decayed
  • A trace identifier, so downstream use can be audited

The contract is designed to be served over an API and over MCP, so Padanet’s context appears inside the assistants and agents you already run. Agents may propose back into the memory; nothing they propose becomes organizational record without governance. The register below says what exists today, what is planned, and what is later — nothing here implies a working connector that does not exist.

Today

Chat where work already happens

Slack, and Padanet's own conversational surface. Other chat vendors are on the roadmap.

Today

Signals from your development tools

Probes on tools like Cursor, Claude Code and git emit observations into the memory. This direction — collecting — works.

Today

Padanet's own apps

The mobile and web apps read and write the memory directly.

Planned

The context packet over MCP

Your agent requesting context for a declared purpose and receiving a packet — recalled against a principal's memory, under Padanet auth and scope. Designed; not available today.

Planned

The context API for your systems

The same request-in, packet-out contract for third-party systems to build against. Retrieval exists for our own agents; the public contract does not yet.

Planned

IDE hooks that query the memory

Asking the memory a question from inside your editor, rather than only feeding it.

Later

A memory-aware model gateway

Routing your model traffic through the memory. Under consideration, not scheduled.

If you need one of the planned paths

Design partners shape what gets built first. Book a call and tell us which one you would actually use.