What is happening?
Keep a current, traceable picture of the work and what changed.
Technology
Connect the reasons.
Control the action.
Verify the effect.
A proposed AI work control plane that keeps delegation, state, proposals, authority, and observed effects connected across the enterprise’s existing systems.
Consequential Action Continuity
Keep your stack.
Connect the conditions.
Before AI changes something real, make sure the conditions that justified the action are still true. After it acts, verify what actually happened.
Do the reasons still hold?
Check that the proposed action remains supported by current state, scope, evidence, and authority.
Did the intended change happen?
Compare the observed result with the intended effect, separately from a tool’s success message.
YOUR IDENTITY · YOUR POLICIES · YOUR WORKFLOWS · YOUR AGENTS · YOUR BUSINESS SYSTEMS
Explore how they connect ↗Enterprise integration direction
A control plane.
A separate operator console.
Latchboard contributes the human/agent interaction model. The generalized control machinery sits beneath it, alongside the systems the enterprise already uses.
The company’s existing systems
CRM · email · ERP · tickets · agents · APIs · databases
OBSERVE
What changed?
PortfolioControl-derived control loopSTATE
What is officially true?
PortfolioControl-derived control loopDELEGATE
What may this AI work on?
Latchboard-derived interaction modelPROPOSE
What change does it want to make?
Latchboard-derived interaction modelCHECK
Is it fresh, on track, and authorized?
Orphard trajectory + DVS release authorityEXECUTE
Make the allowed change.
The enterprise’s existing systemsVERIFY
Did reality change as intended?
PortfolioControl-derived independent checkingObserved results return to official state. Unresolved outcomes remain unresolved.
Inspect the separation and integration scope
PortfolioControl contributes observation, controlled official-state updates, and separate execution/result checking. Latchboard contributes explicit delegation, bounded proposals, and the operator interface. DVS governs release authority; Orphard governs trajectory and recovery.
One logical authority makes work state official; that does not require one server or one process. Observation may inform a decision without acquiring authority to commit it.
Approval requirements depend on policy; not every action needs a human click. The effect check must report uncertainty when the intended result cannot be observed.
This is an enterprise architecture direction to integrate and test. It does not claim that a generalized core, enterprise connectors, or the complete combined contract have already been validated.
View the integration evidence state01 / RELEASE AUTHORITY
DVS
May this exact action happen now?
Keep the proposed transition tied to current authority, evidence, target, policy, and state.Explore DVS 02 / TRAJECTORY & RECOVERYOrphard
Is this still the right work to continue?
Keep long-running work inside an evidenced operating envelope as conditions change.Explore OrphardDVS governs release authority. Orphard governs trajectory and recovery. Reliability cannot create permission, and permission cannot guarantee a reliable trajectory.
System Map
Different jobs.
One learning loop.
Open any node to inspect its role. These are methodological relationships, not transfers of validation or authority.
Crimson Delta → Control R&D → DVS and Orphard → PortfolioControl → Observed failure → Crimson Delta.
01Crimson Delta
- What it does
- Research questions and separately governed programs.
- What it does not do
- Transfer validation between programs.
- Maturity
- OPEN R&D
- Public evidence
- Inspect evidence record
- Latest public result
- Public research structure defined.
- Technical page
- Explore Crimson Delta
02Control R&D
- What it does
- Turn control problems into testable architecture questions.
- What it does not do
- Treat a design relationship as proof.
- Maturity
- ACTIVE R&D
- Public evidence
- Inspect evidence record
- Latest public result
- Comparative integration remains an open question.
- Technical page
- Explore Control R&D
03DVS
- What it does
- Govern release authority for an exact consequential transition.
- What it does not do
- Guarantee trajectory reliability or successful downstream effect.
- Maturity
- LOCAL-ALPHA / PRE-PILOT
- Public evidence
- Inspect evidence record
- Latest public result
- Selected allowed and blocked paths recorded in-house.
- Technical page
- Explore DVS
04Orphard
- What it does
- Govern trajectory and recovery as work changes.
- What it does not do
- Create permission to act.
- Maturity
- ACTIVE R&D
- Public evidence
- Inspect evidence record
- Latest public result
- Public architecture defined; comparative outcomes remain open.
- Technical page
- Explore Orphard
05PortfolioControl
- What it does
- Provide an operating proving environment for control questions.
- What it does not do
- Stand in for customer validation.
- Maturity
- OPERATING PROVING ENVIRONMENT
- Public evidence
- Inspect evidence record
- Latest public result
- Public scope defined; private operating state is not exposed.
- Technical page
- Explore PortfolioControl
06Observed failure
- What it does
- Return an observed problem to research as a question.
- What it does not do
- Automatically validate the proposed repair.
- Maturity
- METHOD
- Public evidence
- Inspect evidence record
- Latest public result
- This map makes no new empirical failure claim.
- Technical page
- Explore Observed failure
An observed failure returns a question to research. A proposed repair still needs its own evidence.
PortfolioControl
Put control questions into the context of continuing work.
Reference demonstrationLatchboard
Inspect a concrete review workflow for AI-proposed changes.
Build with unFragged
Bring one workflow. Build the evidence together.
Establish the strongest competent baseline. Integrate where a residual control gap remains. Measure the delta. Keep a clean negative result if the existing stack already solves it.
Partner with unFragged