CAVO / AI Spend Intelligence
AI costs and measured business value
Public overview pendingMeasurable performance, quality, risk, accountability and evidence for AI-enabled work.
← Research wheelCurrent projects, research tracks and proposals in this area. Listing does not imply production verification.
AI costs and measured business value
Public overview pendingCode quality and testing workflow
Public overview pendingGoverned AI execution research
Explore ↗Bounded, authorized machine actions
Explore ↗Execution cost, quality, capacity and economics
Explore ↗Evidence about cost and capacity
Explore ↗Contracts at machine action boundaries
Explore ↗Measurement of meaningful human control
Explore ↗Authority and human agency
Explore ↗Contract and invoice recovery research
Public overview pendingResearch into conformance and authority
Public overview pendingAutomated verification of software quality
Public overview pendingDeterministic decision recording and execution controls
Public overview pendingAdmissibility and authority for execution
Explore ↗Research reports and evidence
Explore ↗Governed outcome production, risk ownership and evidence-led settlement
Explore ↗Insurance and underwriting demonstrations
Explore ↗Intelligent evaluation in machine action
Public overview pendingExecution evidence and receipts
Public overview pendingRegistered research, applications and technical components in this area. Archives and reference forks are available in the complete inventory.
public HEIMEL authority-to-consequence core: specification, schemas, formal models, HEIMEL-CONFORMANCE/2.0 vectors, minimal reference adapter and signed release attestation (v0.2.0); not the private HEIMEL Enterprise implementation or the VALO technology/product platform runtime
governed world-state owner and deterministic event/replay substrate
typed versioned governed Function substrate
frozen reference implementation of runtime evaluation and governance signals; possible integration source; evaluation runtime owner not yet identified
sole canonical execution-authorization runtime boundary
deterministic decision and interchange contract
mechanical execution enforcement and bounded dispatch
append-only WORM receipt custody and attestation
selection assurance, governed-workspace/state-admission implementation, GCoP protocol-contract profiles, product orchestration and integration compatibility surface
Work OS runtime and governed operator API
MCP integration and governed tool-access distribution surface
supporting deterministic L1 Guardian and formal-verification implementation lineage
core freeze records plus canonical VALO/HEIMEL naming (2026-09-28) and user-time priority invariant (2026-10-10); governance record only, no runtime ownership
Factory OS and software production control plane
research and evidence specialist factory
canonical skill metadata, versions, dependencies and risk contracts
audience-specific governed execution demonstrations and portal build surface
Factory command/policy plane and approved-runner control surface
REHT-specific public-standard and conformance source
portable vendor-neutral governed agent contract and A2A reference implementation; canonical mapping to GCoP remains unresolved
governed agent marketplace and portable reputation layer consuming verified outcomes
outcome-based settlement and reputation effects from verified receipts
active research workspace for systemic agentic execution risk and the missing boundary between workflow completion and execution
brand assets, diagrams, presentation material and communication components; downstream of canonical architecture
reht/demo web-estate implementation surface and canonical demo route estate
shared build and deployment infrastructure for approved VALO/reht web artifacts
bounded comparison harness for AbcadaCLV and REHT against identical execution-boundary scenarios
authority-state and action attestation adapter support; normalizes evidence and never grants execution authority
receipt-based incident reconstruction, causal analysis and root-cause assurance tooling
Distributed Influence Observatory canonical evidence-first implementation for observable influence relationships
source-first genealogy application for identity resolution, evidence, family trees and publication
canonical idea/mission lifecycle, governed intake and portfolio-to-factory experimentation spine
model lifecycle, provenance, MAL and receipt API supporting Model Factory governance
bounded REHT merge-gate pilot and security-evidence experiment; no independent merge or authorization authority
external read-only GitHub event bridge for REHT engineering shadow pilot; normalizes observations and makes no governance decisions
frozen shadow-mode engineering admissibility pilot; never executes or blocks and awaits deployment verification
governance-bypass and unsupported-claim scanner for codebase assurance
multimodal voice/camera Personal OS application profile consuming existing Lens/VAIG/REHT/RACS/Gateway/Veritas boundaries
frozen machine-underwritable assurance domain pack binding AssuranceProfile, commit assurance and claims evidence to the canonical execution chain
carrier-facing machine-underwritable assurance pilot over frozen valo-insurance-pack v0.1.0; does not extend core
VALO-side bounded adapter/profile for external Aurora-Lens governed-state admission; external IP remains independent
canonical SI capability identity, taxonomy, verification, matching and deployment-candidate semantics
governed outcome-production and economic lifecycle after qualified Need handoff, including capability composition, risk/evidence closure, GCU progression and settlement
Verified Outcome-Centric Architecture reconstruction/map connecting distinct outcome, capability, production, continuity, evidence and authority domains without absorbing ownership
EXUP product and commercial contract home: workload improvement against a locked baseline, paid only for verified economic improvement
private EXUP evaluation, specialization and execution workspace with cloud execution policy and evidence capture; next multi-axis proof locked 2026-10-10 but not yet executed
governed control plane and MCP surface for local and remote machine effects through claimed workers and signed worker permits; carries the tool-commander (tc*) entry points
CUA Continuous Underwriting Assurance: independent evidence-based attestation that operation stays inside the risk envelope accepted by an underwriter
independent assurance and red-team program that tries to falsify the governed-effect invariants (no direct effect path, consequence-time authority, fail-closed, exact decision-effect binding)
CAI Central Agent Intelligence: persistent mission-driven discovery, observation and governed action service composing SCOUT, BARO, optional JEV evaluation and HEIMEL
EMOS Enterprise Flight Simulator and CEO game: deterministic falsifiable benchmark comparing unrestricted AI, classic governance and EMOS+ACE
private OLAV voice-selection experiment artifact (single static voice-audition page); Olav is private and outside Heimel
private OLAV voice-selection experiment artifact (single static voice-audition page, inspected 2026-10-11); Olav is private and outside Heimel
private OLAV voice-selection experiment artifact (single static voice-audition page, inspected 2026-10-11); Olav is private and outside Heimel
private OLAV voice-selection experiment artifact (single static voice-audition page, inspected 2026-10-11); Olav is private and outside Heimel
private OLAV voice-selection experiment artifact (single static voice-audition page, inspected 2026-10-11); Olav is private and outside Heimel
private OLAV voice-selection experiment artifact (single static voice-audition page, inspected 2026-10-11); Olav is private and outside Heimel
fork of extend-hq/jevbox (permission-aware document library) carrying own VALO document-hub work: first-party VALO parser adapter, CAI local retrieval and provenance model (8 own commits, merged 2026-10-05); its docs/valo-integration.txt declares the goal of making it VALO's canonical document, report and evidence hub
independent retrospective measurement of observable human review/control; no authorization
single-host agent mandate/outfit/room prototype with scoped rights and observed routing; not canonical REHT or full-path enforcement
candidate private HEIMEL Enterprise delivery runtime with Rust permit/replay/verifier tests; production effects disabled
Archived or incomplete repository; role not verified
Project repository; public description not yet verified
Public project; role unverified
Independent pull-request verification experiments
Experimental research project; public description not verified
Verification workspace research
AI spend, allocation and customer-attested value intelligence
Public educational economics and money simulation
Independent pull-request verification experiments