Details · development papers · 12 November 2026
Unfold what we are building.
Private conversations. A calmer workspace. Communication that can recover from interruptions. Explore the intended experience, then open the engineering detail beneath it.
These are version 0.1 design whitepapers. They describe development direction and acceptance criteria, not certified capabilities or general availability. Each product has its own release decision.
Whitepaper 01 · sha.chatTwo worlds, yours by designPersonal and Professional communication
Abstract. sha.chat is developing an Android communications experience that separates Personal and Professional contexts. The intended value is straightforward: keep private relationships and work conversations organised without allowing one context to silently take over the other. Encryption, identity, recovery and device behaviour must work together before a closed beta is accepted.
01 · Who it is for and what changes
For people balancing personal relationships and professional responsibilities, the proposed experience provides an explicit context before invitations, conversations and professional actions. A collaborator should be able to recognise where an interaction belongs and what information will be shared.
- Personal: invite a trusted contact and test a private conversation on supported Android devices.
- Professional: explore an explicitly separate work context, with professional identity and permissions visible at the point of use.
- Design partners: evaluate a defined workflow with a small invited cohort, report friction and compare improvements against an agreed baseline.
These are intended test journeys. A website demonstration does not establish that every journey is implemented in the Android client.
02 · Architecture and trust boundaries
The Android client, identity and invitation services, delivery infrastructure and device storage form different trust boundaries. The client is responsible for presenting the selected context; backend separation must also be enforced so that an interface switch cannot become permission to cross contexts.
End-to-end encryption is a release requirement. Cryptographic integration must use reviewed protocols and libraries, with versioned profiles and interoperability evidence. Algorithm names alone are not proof of a secure messaging implementation. Professional control signatures and chat encryption are separate functions and must not be described interchangeably.
The browser is not automatically a second chat device. Content access, long-term keys, device linking and Android interoperability require their own approved scope and tests. ZoZ integration must not silently broaden that scope.
03 · Privacy and failure behaviour
Invitations should avoid requiring bulk address-book upload. Context isolation must be tested in both directions, including denied actions and recovery. Failure to establish identity or encryption should produce a visible failure rather than an unannounced weaker path.
Recovery, multi-device behaviour, notification privacy and call routing need separate tests. Screenshots, compromised devices and recipients copying information remain distinct threats; no user interface can promise universal prevention of disclosure.
KAoMe voice and singing features belong to their own project. They are not part of the sha.chat offer.
04 · Delivery gates and evidence
| Gate | Evidence needed |
|---|---|
| Scope and parity | Map advertised options to Android behaviour; identify omissions explicitly. |
| Identity and security | Invitation, authentication, context isolation, encryption and recovery checks on the candidate build. |
| Closed beta | Invited-device journeys, including supported Samsung and Pixel test devices, with failures recorded. |
| Broader release | Resolve critical failures and assess feedback, operational readiness and support. |
The founder target is closed beta by 28 February 2027, subject to the gates above. Opus’s earlier proposal—December scope freeze, February closed test, Q1 scoped comparison and a later standard release—is retained as planning input. Opus proposed the gates; “Opus” is not the project name. Dates do not mark a gate as passed.
05 · What a design partner can evaluate
Start with a small workflow: invite, establish identity, exchange a message in the intended context, restart the device and verify expected access. Measure completion, misunderstanding of context and failure recovery. Extend to calls only when the relevant implementation is ready.
We welcome interest from people willing to test a bounded workflow and provide specific feedback. No payment, security certification or access date is implied by registration.
Register interest in sha.chat →
Whitepaper 02 · ZoZ + ZnetA workspace now. Resilience by evidence.Browser workspace and distributed communication research
Abstract. ZoZ is a browser workspace under development; Znet is a separate distributed-network research effort. The useful first step is a limited workspace people can open and evaluate. Reliable offline operation, encrypted communications and disruption recovery are additional capabilities with different dependencies. Releasing a shell does not prove the network.
01 · A smaller useful first release
ZoZ can first offer a focused starting point for approved tools, project navigation and local workspace interactions. A private QA cohort can assess usability and access controls before any sha.chat client or Znet transport is integrated.
- Independent builder: reach a project’s tools and documentation through a consistent workspace.
- Small team: test approved access from explicitly enrolled personal or professional devices, keeping contexts distinct.
- Interrupted connection: inspect which cached workspace functions remain usable and which require reconnection.
Device enrolment and Personal/Professional containers are requirements to verify, not claims of current universal enforcement. A browser alone cannot reliably identify every employer-managed device.
02 · Offline has several meanings
| Capability | Dependency and boundary |
|---|---|
| Open a cached shell | Prior successful installation/cache and supported browser behaviour. Does not imply sending a message. |
| Queue a local action | Durable local storage, visible pending state, cancellation and replay rules. |
| Reach a nearby peer | A supported physical or network link, compatible peers and explicit permissions. |
| Reach a remote recipient | A verified route or later store-and-forward opportunity. Distance is not bridged by an LLM. |
Internet-free transport depends on the actual radios, links and operating-system/browser capabilities available. Browser restrictions, permissions, battery use and relay availability must be measured. No global delivery guarantee is offered.
03 · Nodes, identity and message handling
The proposed Znet model treats nodes as participants with explicit identity, permitted actions and observable delivery state. LLM assistance may help inspect or route authorised work; it does not replace authentication, cryptography or a physical transport.
Before a network test, define the supported transport, packet limits, expiry, duplicate handling, acknowledgements and recovery after restart. Pending, relayed and recipient-confirmed states must remain distinguishable. A relay accepting an item is not proof that its recipient received it.
Keys and forensic material must remain within the signing user’s authorised scope. Integrations with Z<>W or sha.chat require independent interface and security review; a shared brand does not imply a shared trust boundary.
04 · Release order and revised planning
- QA shell: verify navigation, access controls and explicit feature status.
- Offline workspace: test installation, cache behaviour, restart, storage failure and recovery.
- Private feedback: evaluate the limited workspace with invited testers.
- Communications integration: verify identity, device linking and encrypted interoperability independently.
- Znet transport experiment: prove one supported link and failure/recovery model before broader claims.
Opus’s earlier ZoZ proposal—Q1 2027 proof, Q2 closed test, Q3 scoped release—is preserved as historical planning input. A smaller QA shell may be achievable earlier; its date is not confirmed. Earlier shell delivery does not bring forward the unverified network capabilities. The earlier Znet Q4 2027 proposal is also not an accepted delivery guarantee.
05 · SOS use cases and their limits
Long outages motivate the research: preserve a local draft, communicate over an available supported link, or carry an encrypted item until a route becomes available. Each scenario needs practical testing under loss, delay, restart and depleted-power conditions.
ZoZ and Znet are not currently validated emergency services. They must not be relied on as the sole means of getting help. Emergency-service connectivity, location disclosure and guaranteed delivery are not established features.
We are seeking design partners for bounded workspace and connectivity tests, starting with realistic equipment and a clearly described link.
Explore ZoZ testing → · Register Znet research interest →Help shape a useful first release.
Choose the project that matches a real problem. Tell us where the workflow breaks today. We will use that interest to plan focused testing—not treat it as a sale.
Register project interest →