Six use cases where adopting the Open Knowledge Format gets an organisation to a point of real value — and where the high-value outcome lives in the layer the format deliberately leaves open. Then: who inside those organisations to talk to, and how to make the case.
What this document does. It gives a sales and positioning team a repeatable way to approach organisations that have adopted, or are adopting, the Open Knowledge Format. It names six use cases where the format earns real value and Loomworks is the natural extension to a higher-value outcome, then maps who to present to and how, and consolidates the value propositions those buyers tend not to have thought of yet. It builds on the comparison investigation v0.1 and the two use-case documents.
The through-line. Every use case follows one shape: the format captures the knowledge cheaply and portably, which is real and which Loomworks does not replace; the high-value outcome lives past where the format stops; and Loomworks is the governed, provenance-bearing layer that reaches it. The pitch is never “a better format.” It is “the layer that makes the knowledge you already captured safe to act on, and answerable under consequence.”
What decision is needed. None. This is a positioning aid, not a commitment.
Scope and honesty flag. Consuming and emitting the Open Knowledge Format are recorded candidate directions from the investigation, not shipped Loomworks features. This document positions what the architecture is built to enable; it does not claim that interoperability ships today. Illustrative throughout; no financial, legal, or regulatory advice.
Capture is the on-ramp. The value is past the edge.
The Open Knowledge Format does one thing very well: it lets an organisation capture scattered, tribal, and machine-readable knowledge into a portable, version-controlled catalogue that humans and agents can both read, with no software development kit and no lock-in. For most organisations that is the first time their knowledge has ever been in one legible shape. That value is real, and it is the reason a prospect already has the format in hand when Loomworks arrives.
But the format requires exactly one field of any concept and governs nothing. The moment the captured knowledge has to be trusted, changed, contested, acted on autonomously, or defended to an outside party, the format has no answer — by design. Those five conditions are not edge cases. They are where the high-value outcomes live, and they are exactly what Loomworks supplies. The format is the on-ramp; Loomworks is the destination the on-ramp leads to.
Each names the format's starting point, where it tops out, the Loomworks extension and the outcome it reaches, and the value the buyer has usually not yet considered.
Export the data catalogue — tables, metrics, definitions — into the format so reporting, risk, and filing agents share one context.
The same metric carries different definitions across business lines and jurisdictions, all coexisting silently; a definition change overwrites the old one; nothing records who attested a definition or stops an agent acting on a stale, unapproved one.
Every definition threaded to its attesting office and jurisdiction; changes preserved as corrections; settlement gated by an accountable Operator; attestations anchored through FORAY. The outcome is filing correctly at scale and surviving a regulatory examination without an archaeology project.
Provenance stops being compliance overhead and becomes a defensibility asset. “Show what you reported, under which definition, and why” is answered in minutes from the audit trail — something a competitor running on the bare format cannot produce at all. The cost centre becomes a moat.
The reference enrichment agent crawls decades of code and produces the organisation's first usable catalogue of its estate, in weeks.
A model's inference, a retiring engineer's recollection, and a possibly-wrong old design note all read as the same confident file; nothing grades trust; an autonomous refactoring agent acts on a guess and breaks a system nobody fully maps.
Tie-to-source grading — inferred, recalled, documented, or confirmed against production; agents bounded by OVA so they may not act on inference-only assertions; cross-system dependency composed deliberately in Shaping. The outcome is agent-assisted modernisation that does not break what no one understands.
Knowledge insurance against key-person loss. “What do we know only because one engineer says so, and that engineer retires in March?” becomes a recall query instead of an institutional amnesia event. The estate's trust map doubles as a risk-pricing and due-diligence asset — a gradeable account of what you actually know about your own systems.
Capture findings, datasets, protocols, and prior results as a catalogue for research agents to ground on.
A single case observation and a large meta-analysis read as the same kind of claim; a retracted or superseded result keeps feeding agents silently; the dead ends and disproven hypotheses are discarded entirely.
Evidence threaded to its source and grade; a retraction or supersession propagates through recall to everything that built on it; corrections preserved. The outcome is research that builds only on what currently holds and stops re-running what already failed.
Negative knowledge becomes a first-class asset. Most knowledge bases throw away what was tried and did not work; Loomworks keeps the trajectory. The rejected paths are where research-and-development budget leaks — re-derivation and repeated dead ends. Preserved corrections turn that waste into a queryable asset.
Capture regulations, internal policies, controls, and their mappings as a catalogue for compliance agents.
One rule carries different forms across jurisdictions with nothing to route them; when a regulation changes, there is no way to find what rested on the old rule; the trajectory is lost in an overwrite.
Jurisdiction-scoped assertions, recall on a changed premise, governance over what settles into the compliance posture. The outcome is continuous, defensible compliance across jurisdictions rather than a periodic reconstruction.
Regulatory change becomes a query, not a fire drill. The day a rule changes, you get your exposure surface instantly — every policy, control, and process that depended on the old rule, scoped by jurisdiction. Change-readiness becomes a standing capability instead of a scramble.
Capture equipment, process, supplier, and runbook knowledge as a catalogue for operations and maintenance agents.
Sensor readings, operator observations, vendor specifications, and inferred relationships all flatten into the same concept file; when something fails, there is no way to tell observed from asserted from guessed.
Provenance threads distinguish sensor-origin, human-observed, vendor-asserted, and inferred; governance bounds what agents act on. The outcome is predictive and autonomous operations on knowledge that can actually be trusted, with root-cause analysis that holds up.
The knowledge base becomes the post-incident witness. When something goes wrong, the provenance trail shows exactly what was believed, on what basis, and when it changed — the difference between a defensible root-cause analysis and a guess. Operational memory becomes evidence.
A firm captures its playbooks, frameworks, and ways of working as a catalogue for its own internal agents.
The method is just files; there is no way to bind it to a client engagement, govern how it is applied, distinguish the firm's proprietary way of working from generic content, or sell it as anything.
Method as product — a practitioner's encoded way of working, composed and sold as a unit, carrying an inert contract proposal and driving both an interaction surface and a production surface. The outcome is tacit expertise turned into a delegable, governed, repeatable capability.
Their knowledge stops being internal context and becomes a product. The method is licensable, provenance-backed, and a switching-cost moat — clients who run on the firm's method do not casually leave. For a services firm, this is the move from selling hours to selling an asset.
The team that adopted the format opens the door. The person who owns the consequence signs.
Inside an organisation that has adopted the Open Knowledge Format, four roles matter, and they are not the same person. Approaching the wrong one wastes the meeting.
| Role | What they own, and why they care |
|---|---|
| The technical champion data or artificial-intelligence platform team that adopted the format | Feels the limits first — colliding concepts, ungraded inference, agents acting on bad knowledge. Cannot buy, but opens the door and validates the problem. Win them with the architecture; they carry it inward. |
| The consequence owner Chief Data, Risk, or Compliance Officer; VP Engineering; Head of Research | Owns the downside Loomworks prevents — the mis-filing, the broken core system, the failed audit, the research built on a retraction. This is the economic buyer. The pitch is the cost of the consequence, not the elegance of the layer. |
| The agent-platform owner Head of artificial intelligence or autonomous systems | Is racing to point autonomous agents at the organisation's knowledge and owns the question “can I trust agents to act on this?” That question is the governance gate. In 2026 this is often the sharpest, most current entry point. |
| The regulated-function owner regulatory, quality, safety, or audit lead | Carries the pain in regulated and safety-critical cases. Cares about provenance and the audit trail as survival, not overhead. A powerful internal sponsor once shown that provenance is a defensibility asset. |
Their own adoption of the format is the on-ramp, not the competitor.
The positioning is a build-on, never a rip-and-replace. The prospect invested in the format; that investment is the reason they are a good fit, and the pitch should honour it.
Keep your Open Knowledge Format. Loomworks reads it and writes it back, so your agents keep speaking the format they already read. We are not a new catalogue. We are the layer that makes the knowledge you already captured safe to act on — and answerable when someone asks you to prove it.
One question opens every one of the six use cases, because the format has no answer to it:
What happens when that knowledge is wrong, changes, is disputed, or an autonomous agent has to act on it — and someone outside has to be able to trust the result?
Let the prospect sit with it. The catalogue they are proud of has no response. That silence is the opening, and it is honest — it is the format being exactly as minimal as it promised to be, not a flaw to attack.
Land on one consequential engagement — the regulated metric, the legacy subsystem nobody understands, the body of policy under a jurisdiction in flux — where the cost of the knowledge being wrong is concrete and a person owns it. Prove the governed layer there. Then expand across the estate, one engagement at a time. The format already covers the breadth; Loomworks earns its way in through depth where the consequence is sharpest.
In 2026 the agent-safety angle lands hardest. Everyone is racing to point autonomous agents at their knowledge. The format feeds those agents confident knowledge with no grade on it — an unverified guess and a production-confirmed fact look identical. Loomworks is the provenance and governance that makes agent action on that knowledge trustworthy. To an organisation about to delegate real decisions to agents, that is not a feature; it is the precondition.
The reframes that turn a “nice layer” into a must-have. These are cross-cutting; pull the ones that fit the room.
The audit trail most organisations treat as a compliance cost becomes an asset they can show a regulator, a board, or an acquirer on demand — and one their bare-format competitors cannot produce.
When a definition, rule, or premise changes, recall surfaces the full exposure surface instantly — everything that rested on it. The fire drill becomes a search.
Preserved corrections keep the dead ends, the disproven, the superseded. Most systems discard exactly the knowledge that stops an organisation from re-deriving and re-failing.
Autonomous agents acting on ungraded knowledge is the risk no one has fully priced. The governance gate and the trust grading are what make delegation to agents survivable.
Because every assertion ties to its source, “what do we know only because this one person says so” is answerable — before they leave, not after.
Shaping arranges the same Memory for a regulator, an agent, a customer, and a board — each with the right depth of provenance, and no drifting copies to reconcile.
Loomworks emits the Open Knowledge Format back out. The prospect keeps the format their agents and tools already speak; the governance and trajectory live on the Loomworks side. Adoption is additive and reversible.
For firms whose expertise is the business, a governed, provenance-backed method becomes something licensable and defensible — a switching-cost moat, not just internal context.
Positioning aid, internal. All scenarios illustrative; no financial, legal, regulatory, or engineering advice.
Open Knowledge Format behaviour drawn from the v0.1 specification and the Google Cloud announcement of 12 June 2026 (concept-per-file, type-only requirement, reference enrichment agent, producer/consumer independence, no origin or trust grading, change recorded as overwrite plus timestamp).
Loomworks behaviour drawn from the candidate seed (four rooms; Shaping arranges Manifestation-organised knowledge for a specific reader; references between engagements non-transitive by default; composition in Shaping; source-supplies/Operator-governs; Operator-authority over state transitions), the provenance-threads framing (outward threads to person, document, sensor, query, scope; inward-inference thread; tie-to-source trust axis; non-suppressible machine-origin marking), OVA authorisation reach, FORAY attestation and anchoring, the method-as-product line, and the three prior Open Knowledge Format documents (comparison investigation v0.1, restatement use case v0.1, legacy-estate use case v0.1).