Linne Engine is a shared foundation for intelligent media, realtime production, rendering, spatial systems, and evidence-led creative tools.
204
Tracked roadmap items
126
Proven at stated scope
9
Product-integrated
12
Public system groups
Latest progress · July 13–17
This week focused on making the Engine more autonomous, more portable across machines, and more deeply qualified in the native workflows that production depends on.
The first-human work now retains a coarse candidate and measures the same geometry in its inspection report. The candidate remains rejected for production use while visual review and later fit work remain open.
Published FH-05–FH-12 task receipts and focused engine candidate-inspection proof
Authenticated social and personal archive sources now carry author, removal, and replay boundaries into corpus processing without treating fetched content as automatically eligible training evidence.
Engine archive-intake and social-composition changes published in September 2026
The engine can diagnose and recover bounded Broker service failures and report developer-storage ownership with retained proof of the packaging boundary.
Merged Broker recovery and developer-storage attribution proof changes
Removed training exchanges no longer reappear in retained sessions, and partner continuations fail closed when their selected evidence is invalid.
Merged September 2026 chat-twin removal, selection, and continuation changes
Capability map
Browse the public capability projection. Internal implementation details, private operational records, and customer-specific systems are intentionally excluded.
Showing 12 of 12 system groups
12 systems · 70 system links · 334 cross-system feature dependencies · snapshot 2026-09-02
Arrows run from a capability dependency to the system that uses it. Line weight reflects the number of public feature relationships—not runtime traffic or implementation coupling.
204
Public features
628
Feature links
5
Curated paths
Lifecycle & orchestration
A shared control layer for starting, discovering, attaching to, supervising, and coordinating production workloads across machines.
Example proof: Integrated peer recovery, revision handoff, and Receiver workflows
View all featuresGraphs, scenes & spaces
Structured creative intent remains separate from live runtime state, making workflows portable, inspectable, and safe to evolve.
Example proof: Repeatable authoring and command proofs
View all featuresAdmission & placement
Hardware capability, realtime pressure, budgets, and policy are evaluated before work is admitted or placed.
Example proof: Deterministic planning and refusal proofs
View all featuresProvider-neutral execution
Applications request a capability instead of hard-coding a model, while execution, evidence, and recommendations stay inspectable.
Example proof: Local and distributed execution proofs
View all featuresIdentity, analysis & generation
A durable media layer joins analysis, detected subjects, depth, speech, captions, image generation, and responsible source use.
Example proof: Artifact, command, and scenario proofs
View all featuresControllable creative systems
Executable generation workflows support still images, scene extension, layered audio, verification, provenance, and runtime control.
Example proof: Engine-level execution and artifact verification
View all features2D, 3D & broadcast graphics
A shared rendering platform covers photoreal output, deterministic motion graphics, real font geometry, portable packages, and distributed work.
Example proof: Product, render, package, and distributed proofs
View all featuresRealtime production flow
Native-first audio, video, streaming, protocol qualification, remote endpoints, and recovery form one observable production path.
Example proof: Hosted protocol, Receiver, cadence, and recovery qualification
View all featuresRooms, devices & calibration
Spatial truth models rooms, scans, devices, calibration, and media placement without mixing authored intent with live measurements.
Example proof: Engine-level spatial and device verification
View all featuresEvidence-led digital humans
Source-derived bodies, faces, motion, materials, garments, identity evidence, and consent are assembled with explicit unknown and blocked states.
Example proof: Native biomechanics workflows proven; final visual and physical qualification remains open
View all featuresEditable, explainable experiences
Structured pages, editable content, contextual help, feature knowledge, interaction facts, and source-aware explanations share one model.
Example proof: Engine-level content and knowledge verification
View all featuresSecurity, privacy & validation
Cryptography, consent, retention, evidence, deterministic scenarios, and fail-closed policies are designed into capability delivery.
Example proof: Focused security, policy, and scenario proofs
View all featuresHow to read this
A capability can be real in the engine without being generally available in a product. Each item carries its roadmap stage or bounded proof level; these labels are not release dates, service guarantees, or claims of universal quality.
View structured catalogAt least one slice in this system has first-party product use; other items can remain planned.
At least one slice has bounded engine proof; broader qualification and other items may remain open.
At least one slice has command, scenario, or repeatable proof; complete system availability is not implied.
The system includes implementation or design work with qualification still open.
The catalog will evolve as capabilities move from engine proof into focused products and production environments.