On this page
apu.bot · Order & defend
Make incoming work safer to use.
Turn incoming material into an outline you can inspect, revise and keep using. A source, a tentative idea and an accepted decision should not become the same thing.
One job, with clear boundaries.
- Starts with
- Material you deliberately address to it, with the account, permission and source revision attached.
- Produces
- Source-linked structure, bounded reasoning, explicit conflicts and questions where intent is unclear.
apu.bot is independent of Menexus. It does not silently retain every conversation, publish sources or turn a suggested next step into permission to act.
One useful interaction.
- Bring a maker’s specification and a compatibility question.
- Separate the original facts, uncertain fit and possible next steps.
- Review a reversible outline; retain or revise it in the chosen account.
The test should include missing evidence, a changed source and a person declining—not only a successful example. Compare usefulness, effort and cost with the existing process.
Direction and delivery · VMOSK-A
A readable overview of the source-owned delivery direction. Vision and mission set the direction; objectives, strategies, measures and skills make the work testable.
- Vision
- Understanding remains useful after a conversation ends, without losing uncertainty or control.
- Mission
- Organise permitted sources and evolving ideas into an inspectable outline with reversible revisions.
- Objectives
- Prove stable outline identities, faithful tentative notes, exact undo and account isolation before expanding integrations.
- Strategies
- Start with source-preserving retrieval and a simple baseline. Adopt more elaborate reasoning only when it improves a declared task.
- Measures & triggers
- Measure unsupported answers, source coverage, stale cross-account writes, revision fidelity, latency and total cost separately.
- Agents & skills
- Interaction, memory-semantics, retrieval, security and evaluation skills are required. Delivery staffing and owners remain to be confirmed.
Plan basis and open decisions
This overview does not replace or promote the owning project’s VMOSK-A. Delivery staffing, current KPI baselines and numerical acceptance thresholds require their own source decisions and measurements.
- apu.bot delivery plan · 2026-09-24
- Reasoning-facet account · 2026-08-19
Distinct responsibilities.
Its L1 intake and L5 reasoning facets belong to one utility. Neither makes it the Menexus AIA or the owner of another app’s decisions.
- Applied L1
- Intake, ordering and defence
- Applied L5
- Reasoning over a permitted view
How the Rosetta correspondence is used
These are documented applied functions, not a ranking of people or a seven-products-to-seven-rows assignment. Their correspondence to Rosetta is an interpretive projection. Dispatch stations review work; they are not new app identities.
The source map preserves the dated applied namespace while its selected successor awaits a separately reviewed migration. Dispatch L5–L7 are counsel, not autonomous execution modes; applied product functions are a different register. A page, plan or model may neither sign nor authorise an action.
Source basis: September 2 level reconciliation, corrected product taxonomy and September 14 owner-selected allocation. Transport, exchange applications and capital products do not acquire unused seats by their position in a catalogue.
Explore the design.
These pages describe the intended product and its delivery work. They do not open a signed-in app or establish operating availability.