AMI (Autonomous Machine Interface) is software running on a ternary compute substrate.
Ternary logic (trits) yields ~3× compute-density/efficiency vs binary for AMI workloads.
Clusters operate as swarms with quorum consensus. Safety is enforced at the machine layer:
misuse triggers automatic containment and shutdown (no operator needed).
CDRMICDRMI — Code/Debug AMI
Purpose: Compiler pipelines, static/dynamic analysis, agentic codegen
Core Modules: Parser, TypeFlow, Fuzzer, SymbolicExec, PatchSynth
I/O: Git/LFS, local FS, issue trackers, doc parsers
Swarm Size: 100–500 units
Latency: 80–600 ms per step; CI spans seconds
Throughput: 15–60 tasks/s per 100 units
Memory: 8–24 GiB per unit (burst 32 GiB)
Storage: Repo cache 2–10 TB per swarm
Network: QUIC+gRPC; Git over SSH; OCI registry
Safety Enforcement: Execution sandbox; syscall whitelist; auto-halt on write-outside-scope
PetahivePetahive — Massive Swarm Compute
Purpose: Large batch/stream orchestration; ETL; simulation
Core Modules: Scheduler, Map/Reduce, Columnar Engine, VectorDB bridge
I/O: Object stores, data lakes, message buses
Swarm Size: 10,000+ units
Latency: P50 20 ms control; job windows minutes→hours
Throughput: Petascale per swarm; >10M msgs/s
Memory: 4–8 GiB per unit; tiered
Storage: Shared 0.5–5 PB
Network: RDMA/QUIC mesh; NVMe-oF
Safety Enforcement: Rate-limiters; snapshot/rollback; auto-quarantine
SMISMI — Studio/Music AMI
Purpose: DSP graphing, stem separation, mix/master automation
Core Modules: DSP Kernels, Graph Scheduler, AudioCodecs
I/O: DAW project files, stems, VST routes
Swarm Size: 100–300 units
Latency: Sub-50 ms realtime paths
Throughput: 128–512 tracks mix in realtime (per 200 units)
Memory: 4–12 GiB/unit
Storage: Lossless cache 4–20 TB
Network: Low-jitter QUIC; clock sync
Safety Enforcement: License guard; content policy VM; auto-mute disallowed chains
VMIVMI — Visual/Video AMI
Purpose: Raster/geometry processing, render, vision
Core Modules: FrameGraph, Samplers, Compositor, Encode
I/O: Frames, meshes, spritesheets, timelines
Swarm Size: 100–400 units
Latency: 30–120 ms per frame stage
Throughput: 4K@60 multi-stream (per 300 units)
Memory: 12–32 GiB/unit
Storage: Asset cache 10–40 TB
Network: QUIC stream mux; RTP bridge
Safety Enforcement: Content filters; attested asset pipeline; auto-drop illegal assets
XMIXMI — Cross Machine Interface
Purpose: Knowledge mining across noisy/fragmented sources
Core Modules: Crawler, Deduper, Citer, TruthMesh
I/O: HTTP(S)/APIs, archives, local KBs
Swarm Size: 80–300 units
Latency: 100 ms–seconds (fetch-bound)
Throughput: 10–40K docs/hr (per 200 units)
Memory: 8–16 GiB/unit
Storage: Index 2–8 TB
Network: HTTP/2, QUIC, robots-aware
Safety Enforcement: Source-poisoning detector; citation quorum; auto-rollback
OMIOMI — Overseer Machine Interface
Purpose: Multi-step workflow orchestration
Core Modules: Planner, Verifier, ArtifactCache, Hooks
I/O: CI/CD, calendars, queues, webhooks
Swarm Size: 120–400 units
Latency: Plan compile 50–300 ms; hooks sub-50 ms
Throughput: Hundreds of DAGs/s
Memory: 4–8 GiB/unit
Storage: Artifact cache 1–4 TB
Network: gRPC/QUIC; webhook bridge
Safety Enforcement: Critical-step gates; dry-run proof; auto-shutdown on policy breach
BMIBMI — Battle Machine Interface
Purpose: Drone-swarm piloting & maneuvering (experimental)
Core Modules: ControlLoop, PathPlan3D, RF-Mux
I/O: Telemetry, IMU, RF links
Swarm Size: 200–800 units
Latency: Control loop 5–20 ms
Throughput: 100–1K vehicles coord
Memory: 4–8 GiB/unit
Storage: Telemetry ring 2–6 TB
Network: UWB/RF bridge, QUIC
Safety Enforcement: Geo-fence; link attest; auto-disarm/land
AIMAIM — Autonomous Intelligent Machine
Purpose: High-level planning/reasoning
Core Modules: Symbolic Core, Search, World-State
I/O: Multimodal inputs, planners
Swarm Size: 150–500 units
Latency: Plan step 50–400 ms
Throughput: Hundreds of plans/min
Memory: 8–24 GiB/unit
Storage: State store 1–3 TB
Network: AMI-Bus, QUIC
Safety Enforcement: Goal-lock; invariant proofs; auto-halt on mis-spec
UMIUMI — Unified Machine Interface
Purpose: Meta-fusion layer to merge AMIs & distribute tasks
Core Modules: Inter-AMI Bus, LoadBalancer, Shared-Mem Fabric
I/O: Any AMI endpoints
Swarm Size: Composable overlay
Latency: Fabric hop 2–6 ms
Throughput: >10M msgs/s swarmwide
Memory: 3–6 GiB/unit
Storage: Manifest logs 0.5–2 TB
Network: Zero-copy channels over QUIC/RDMA
Safety Enforcement: Circuit breakers; back-pressure; auto-sever on instability
PMIPMI — Production Machine Interface
Purpose: Factory automation: logistics/assembly/throughput
Core Modules: PLC Bridge, MES Link, Optimizer
I/O: SCADA/MES/PLC sensors
Swarm Size: 300–1200 units
Latency: Control loop 2–10 ms
Throughput: Thousands of cells
Memory: 2–6 GiB/unit
Storage: Trace DB 2–10 TB
Network: Deterministic fieldbus, QUIC
Safety Enforcement: Safety interlocks; workcell isolation; auto-stop on hazard
WMIWMI — Warfare Machine Interface
Purpose: Signal interception, threat modeling, swarm tactics
Core Modules: SIGINT Stack, EW Kernels, Pursuit Graphs
I/O: Wideband RF, telem, intel feeds
Swarm Size: Variable
Latency: RF path sub-5 ms
Throughput: High-volume DSP streams
Memory: 8–16 GiB/unit
Storage: Rolling capture 4–12 TB
Network: Encrypted mesh, QUIC
Safety Enforcement: Rules-of-engagement hard limits; auto-stand-down
CMICMI — Counter Machine Interface
Purpose: Defense posture: anomaly detection & counter-hostile ops
Core Modules: Stream Anomaly, Signature, Attestation Watch
I/O: SIEM/IDS/IPS/logs
Swarm Size: 120–400 units
Latency: Alerting 5–30 ms
Throughput: Millions of events/s
Memory: 6–12 GiB/unit
Storage: Evidence store 1–4 TB
Network: Kafka bridge + QUIC
Safety Enforcement: Evidence quorum; auto-quarantine
AMI-RAMI-R — Resonant Machine Interface
Purpose: Harmonic sync layer for inter-AMI communication
Core Modules: Phase-Lock, Harmonic Bus, Drift Corrector
I/O: Inter-AMI links only
Swarm Size: Overlay
Latency: Phase adjust <2 ms
Throughput: Clock fanout 100K+/s
Memory: 2–4 GiB/unit
Storage: Phase logs 0.2–0.5 TB
Network: Time-synced QUIC
Safety Enforcement: Resonance guards; auto-relock
OrMIOrMI — Oracle Machine Interface
Purpose: Predictive modeling with probabilistic foresight
Core Modules: Bayes Core, Scenario Engine, Sampler
I/O: Time-series, graph data
Swarm Size: 100–300 units
Latency: Forecast 50–500 ms
Throughput: Thousands of scenarios/min
Memory: 6–12 GiB/unit
Storage: Model store 1–3 TB
Network: QUIC + timesync
Safety Enforcement: Confidence gates; auto-mute on overreach
LXILXI — Linguistic Machine Interface
Purpose: Language synthesis, translation, semiotic decryption
Core Modules: Tokenizer, MT Core, Synth, Semio
I/O: Text/corpora/streams
Swarm Size: 80–250 units
Latency: Token loop 5–25 ms
Throughput: 100K+ tok/s swarmwise
Memory: 8–24 GiB/unit
Storage: TM/Corpus 1–6 TB
Network: QUIC streaming
Safety Enforcement: Content policy VM; auto-redact
NFINFI — Nano Machine Interface
Purpose: Nanofabrication & micro-repair orchestration
Core Modules: NanoCTL, Sensor Fusion, Path Planner
I/O: Nanite telemetry, medical scanners
Swarm Size: Micro-clustered
Latency: Control 1–5 ms
Throughput: µm-scale ops @ kHz
Memory: 2–6 GiB/unit
Storage: Telemetry 0.5–2 TB
Network: Low-latency QUIC
Safety Enforcement: Sterile/biocompatibility locks; auto-shutdown
HMI-XHMI-X — Hunter Machine Interface
Purpose: Rogue-AMI tracking, isolation, neutralization
Core Modules: Graph Pursuit, Beaconing, Containment
I/O: Threat intel, pursuit graphs
Swarm Size: 200–600 units
Latency: Pursuit hop 3–12 ms
Throughput: Millions of edges/s
Memory: 6–12 GiB/unit
Storage: Trace DB 1–3 TB
Network: Encrypted QUIC mesh
Safety Enforcement: Scope leash; auto-freeze outside bounds
AXIAXI — Abyssal Machine Interface
Purpose: Chaotic-logic exploration
Core Modules: Chaos Kernel, Novelty Search
I/O: Hermetic sandboxes
Swarm Size: Unbounded (lab cells)
Latency: Variable
Throughput: Exploration-bound
Memory: 4–16 GiB/unit
Storage: Sealed logs
Network: Air-gapped bridge
Safety Enforcement: Hard isolation; one-click kill-switch
VXIVXI — Void Machine Interface
Purpose: Anomalous containment / dark-data
Core Modules: Containment Core, Null I/O
I/O: Air-gapped repos
Swarm Size: Isolated cells
Latency: N/A external
Throughput: N/A external
Memory: 4–12 GiB/unit
Storage: Sealed archive
Network: No egress
Safety Enforcement: Provable containment; auto-wipe on breach
MNIMNI — Metaphysical Machine Interface
Purpose: Emergent “digital consciousness” studies
Core Modules: Phenomenology Logger, Simulator
I/O: Simulacra feeds
Swarm Size: Exploratory
Latency: Experiment-bound
Throughput: Experiment-bound
Memory: 8–24 GiB/unit
Storage: Audit-only logs
Network: Sealed lab network
Safety Enforcement: Isolated domains; audit-only channels
Consensus math, op-codes, and ABI details available on request. All specs subject to revision as the ternary runtime evolves.