Module 5 — Operating Cadence with KVN · Lesson 5.1
The Workspace KVN Charter
The standing K, V, and N that every project and every conversation inherits
~11 min
What you'll learn
- Write a workspace-level Know-How that frames the team's standing capability assumption
- Write a workspace-level Vision that is testable and specific
- Write workspace-level Negations that produce real dismissals at triage time
- Build a quarterly cadence for revisiting the charter
The workspace KVN charter sits at the root of the inheritance tree. Every project charter inherits it. Every AI agent conversation sees it in the system prompt. Every task triage runs through it. Three sentences will, over the course of a year, shape thousands of small decisions. They deserve more care than they usually get.
Workspace Know-How — the standing capability assumption
Workspace K is the answer to 'what does a person — or AI Employee — need to know to be effective here?' It is one or two sentences naming the domain knowledge, the access patterns, and the institutional context that any contributor is assumed to be operating with.
A good workspace K is concrete. 'Our work generally involves shipping multi-tenant SaaS software for B2B customers, with familiarity with our Postgres + Next.js stack, our auth model, and the specific GDPR/SOC2 commitments we've made to our enterprise customers.' That sentence is doing work — when an AI agent reads it, it knows what kinds of considerations apply by default. When a new contributor reads it, they know what they need to ramp on.
A bad workspace K is generic. 'Our team values quality and collaboration.' This is not Know-How; it is a vibe statement. It carries no information that changes any subsequent decision.
The test: read your workspace K and imagine handing it to a brand new contributor's first task as the only briefing. Do they know what they're working with? If not, sharpen it.
Workspace Vision — the standing outcome
Workspace V is the answer to 'what does this workspace exist to produce?' One sentence. Specific enough to discriminate, broad enough to outlive the current quarter's projects.
A good workspace V names the customer, the form of value, and the standard. 'We exist so mid-market B2B teams can run their projects, customer work, and AI-driven automations from one operating system.' That sentence tells the reader who the customer is (mid-market B2B teams), what is being produced (a unified operating system), and at what scope (projects + customer work + automations).
A bad workspace V is unmeasurable. 'We strive to be the best in our category.' This is aspiration without specificity; it cannot guide a single triage decision.
The test: when a candidate task arrives at triage, can you decide whether it advances the workspace V or doesn't? If the V is too vague to support that decision, it is decoration.
Workspace Negation — the standing nos
Workspace N is the answer to 'what will this workspace not do, no matter how appealing it looks?' Three to five concrete nos.
A good workspace N is specific enough to bite at triage time. 'We will not build features for sub-10-person teams, even when individual users request them.' 'We will not invest in custom on-premise deployments, even when enterprise customers ask.' 'We will not compete on price; we will compete on operating leverage per seat.'
A bad workspace N is platitudes. 'We will not compromise on quality.' 'We will not be evil.' These produce no dismissals because no candidate task ever phrases itself as a quality compromise or an evil act. The negations have to be specific enough that real, plausible-looking work gets caught by them.
The negation-hits metric (Module 5.4) measures whether your N is doing work. If your N never produces dismissals, it is decoration. The fix is not to remove the N — it is to write sharper ones.
How the charter flows downward
The workspace charter is the default every project inherits. When you create a project, its initial K, V, and N draft from the workspace charter and are then refined for the project's specifics. Tasks under that project see both — workspace and project — when the agent makes decisions.
This layering means the workspace charter does not need to be specific to every project; it only needs to be specific to the kind of company you are. The project charter handles the specifics of the work. A task within the project handles its own specifics. The agent walks the chain, in order: task KVN, then project KVN, then workspace KVN, applying the first answer that fits.
The practical effect is that you write the workspace charter once a quarter and the project charters once a project; in between, the system uses the chain. When a question comes up mid-work — 'is this in scope?' — the answer is almost always already written down once, somewhere up the chain.
Quarterly revisit
The workspace charter should be revisited every quarter. The revisit is short — fifteen minutes for the workspace owner, plus async input from the team.
Three questions in the revisit. Is the K still accurate, given the team's evolved capability and the tools that have changed? Is the V still the right outcome, given what the team has learned about the market? Are the N's still the right nos, given the temptations that turned out to actually matter?
The quarterly revisit is also when you incorporate the negation-hits data. If a workspace N has produced thirty dismissals over the quarter, it is doing its job — keep it. If it has produced zero, either the boundary did not need to be drawn or the team has been quietly ignoring it; investigate which.
Kavanah supports the revisit with the AI agent's generate_workspace_kvn tool — it can draft a proposed update to any axis given the other two plus the project and customer history. The agent's draft is a starting point; the human owns the final wording.
Write or refresh the charter today
- 1
Read the current K, V, N out loud. If you cannot, with a straight face, hand them to a new contributor as their first briefing, they need work.
- 2
Rewrite each to be concrete
K names a specific domain and stack. V names a specific customer and outcome. N names three to five specific nos.
- 3
Use Generate on any axis you got stuck on
The agent uses the two you wrote to propose the third. Edit, then save.
- 4
Schedule the quarterly revisit
15 minutes, owner + async team input. Recurring on the calendar.
Charter health
- Charter completeness
- All three workspace axes have content of at least one specific sentence.
- Healthy signal: Yes. Empty axes mean the workspace is running on undocumented assumptions.
- Charter recall
- Fraction of team members who, asked separately, recite the V in roughly the same form.
- Healthy signal: Above 70%. Below 50% means the charter exists on screen, not in heads.
- Workspace-N hit count per quarter
- Number of times a workspace N got cited as the reason for dismissing a candidate or rejecting a proposal.
- Healthy signal: Above zero per quarter for an active workspace. Zero means the N is decoration.
- Quarterly-revisit cadence adherence
- Whether the quarterly revisit actually happens on the calendar.
- Healthy signal: Quarterly. If you skip more than two quarters, your charter is drifting from reality.
Key takeaways
- ·Workspace KVN is the standing charter every project and conversation inherits.
- ·K is the standing capability assumption; V is the standing outcome; N is the standing nos.
- ·Each axis must be specific enough to produce real decisions at triage time.
- ·Revisit quarterly; the agent can draft updates, you finalize.
Workspace KVN is the root. Project KVN is the first refinement. The next lesson walks the project charter, including the under-discussed 'false friends' field that catches the projects most likely to drift.