HUAI · Systems Baseline

A dated before-state.
Not a frontier deployment claim.

The April 2026 HUAI Baseline Company Asset Proposal documents a structured LLM-systems asset base: architecture maps, registries, measurement, protocol discipline, runtime logic and validation/readiness packaging.

Snapshot boundary: figures on this page describe the HUAI workspace captured in that April 2026 baseline. They are not the 330+ Phase 2 portfolio count, not an independent audit and not a current hyperscale-deployment metric.
L0–L8Systems Baseline
ArchitectureMaps · anatomy
RegistriesNodes · endpoints
MeasurementContracts · signals
ProtocolsControls · rollback
RuntimeLab · replay
ValidationVerdicts · readiness
BaselineApril 2026
Phase 3Independent test
Snapshot · issued 18 Apr 2026 (Europe/Paris)

What the baseline recorded.

The proposal was built for reviewer and strategic-partner diligence and intentionally excluded confidential intervention methods and private future-state implementation pathways.

34Packaged HUAI artifacts recorded in the captured workspace.
1,086Packaged files across HUAI zip artifacts in that snapshot.
L0–L8Layered current-LLM systems decomposition.
6Primary asset classes: architecture, registry, measurement, protocol, runtime/infrastructure, validation/readiness.
Interpretation: these counts establish packaging density and a dated systems baseline. They do not by themselves prove novelty, performance, commercial value, production readiness or independent validation.
Asset classes

A systems-asset story, not only a model story.

Architecture assets

Layered system decomposition

L0–L8 anatomy, current-practice framing and endpoint-candidate mapping designed for structured review.

Registry assets

Turn maps into data structures

Unified layer, node, endpoint and component registries plus readiness matrices and manifests.

Measurement assets

Define what should be measured

Signal taxonomy, endpoint instrumentation matrix, baselines and evidence contracts.

Protocol assets

Make experiments reproducible

Control/counterfactual cases, sequencing, abort logic and rollback contracts.

Runtime / infrastructure

Go beyond architecture notes

Recipe/data/compute assumptions, lane topology, sandboxing, replay and runtime safety rails.

Validation / readiness

Package judgment explicitly

Comparative validation grids, verdict taxonomy, cross-bench logic and intervention-readiness packaging.

Current-LLM decomposition

The L0–L8 systems map.

This is a technical baseline model inside HUAI. It is not the same object as the 21-slot LLM Company Anatomy and not the same object as the cross-phase human-context architecture.

L0

Objective

Training objective, curriculum logic, pre/post transition framing.

L1

Representation

Input handling, normalization, segmentation, tokenization, encoding.

L2

Computational Core

Residual stream, attention, FFN/MoE, normalization, layers.

L3

Context / Memory

Windowing, KV carry, compression, retrieval-fed context.

L4

Routing / Activation

Expert routing, sparse activation, tool dispatch, balancing.

L5

Decoding / Emission

Logits, sampling, output gating, emitted fragment.

L6

Runtime / Serving

Prompt assembly, batching, cache, transport, delivery.

L7

Post-training Behavior

Preference shaping, refusal, moderation overlay, repair.

L8

Evaluation / Observability

Telemetry, attribution, adjudication, release and rollback logic.

Structured registries

Documentation maturity is part of the asset—but not the verdict.

The baseline records unified maps, machine-structured node/endpoint registries, manifests and checksums. This makes later deltas more reviewable while keeping validation separate.

Layer map

One coherent stack

A structured view of system layers rather than scattered notes.

Node registry

Machine-structured nodes

Important system nodes represented in a reviewable registry.

Endpoint registry

Candidate endpoints

Architecture-adjacent terminal points structured for comparison and reasoning.

Manifests / hashes

Packaging + provenance

Versioning and checksums support artifact integrity discussion; self-issued integrity is not independent validation.

Build trajectory

The baseline emerged in stages.

Build bandContributionInterpretation
Baseline mappingEarly architecture maps, layered structure, dashboarded contextInitial systems decomposition.
Canonicalization / hardeningReference overlays and canonical packagingPolish and consistency—not external validation.
Pilot operations / adjudicationIntake, blind-review structures, decision/report/release packagesOperational discipline assets.
Current-LLM anatomy / handoffLayer-complete anatomy, synthesis and handoff reviewStructured current-system understanding.
Track B lab / readinessMeasurement, protocol, recipe/data/compute blueprint, runtime lab, endpoint bench, comparative validationReadiness architecture for future interventions/tests.
Internal maturity snapshot

The baseline was strongest in structure and measurement—not mass-scale deployment.

These are the baseline document’s own internal maturity labels. They are useful as a self-assessment snapshot, not an independent rating.

Capability bandBaseline labelInterpretation
Architecture understandingStrongSerious layered decomposition baseline rather than generic narrative.
Structured artifact maturityStrongRegistries, manifests, templates, dashboards and synthesis packs present.
Evaluation / measurement assetsStrongMeasurement, signal taxonomy, validation and safety logic are asset classes.
Runtime / control logicGoodRuntime lab, lane topology, replay and isolation logic structured.
Data / recipe / compute blueprintingGoodDesign assets beyond ad hoc notes; not hyperscale-training proof.
Mass-scale deploymentIntentionally not claimedBaseline is not a frontier deployment announcement.
Explicit non-claims

The baseline is more useful when its boundary is visible.

No hyperscale claim

Not a finished frontier company

The baseline does not announce a mass-scale frontier training or consumer inference operation.

No method disclosure

Confidential mechanics excluded

Private intervention logic, proprietary execution methods and future-state pathways remain outside the public baseline.

No inference by omission

Only explicit claims count

If a capability is not explicitly established in the baseline, it should not be inferred from surrounding documentation density.