Skip to main content
AIfa Cognitive Runtime · ACR 30 Innovations
Commercial website accessibility service (ADA / WCAG)

Looking for a website audit against ADA Title II / WCAG 2.1 AA for municipalities or commercial portals? Go to the AIfaFocus accessibility scanner.

Go to the accessibility scanner →
AIFA COGNITIVE RUNTIME (ACR) · CONNECTOME-GROUNDED COGNITIVE RUNTIME

AIfa Cognitive Runtime (ACR)

A connectome-grounded cognitive runtime for autonomous software agents, derived from the FlyWire v783 Drosophila whole-brain connectome (139,255 neurons, 54.5M synapses). Five biological computation motifs implemented on CPU (Python/NumPy), no GPU; every number on this page comes from an open script.

Founder, Creator & Chief Architect: Maksim Valentinovich Galatin
Reference Substrate: FlyWire FAFB v783 (139,255 neurons · 54.5M synapses)

Reference substrate: FlyWire FAFB v783 — 139,255 neurons, 54.5M synapses. Dual-mode architecture: Faithful Connectome Mode (topological integrity, synapse graphs & SHA-256 hash) · Distilled ACR Mode (CPU algorithms for software agents; the connectome is used for graph statistics).

Ablation Matrix — real measurement (360 episodes per configuration, 23.09.2026)

Agent / configurationDistractorsTask successSteps when solvedWrong clicksGoal lost / clickDecision P50
Naive agent (goal falls out of a 4-message window)35%0.3% ± 1.510.098.6%0.0000.074 ms
Standard agent, goal pinned (system-prompt style)35%5.8% ± 3.85.7197.9%0.0000.076 ms
+ APL novelty gate only35%4.2% ± 4.23.2797.7%0.0000.076 ms
+ CANN focus ring only35%5.6% ± 3.95.4597.9%0.0250.097 ms
+ APL + CANN35%1.4% ± 3.16.2099.2%0.0010.100 ms
+ APL + CANN + CX (as shipped in aifa_sdk)35%1.4% ± 3.16.2099.2%0.0010.181 ms
Standard agent + visited-link memory only35%95.0% ± 5.56.1653.8%0.0000.218 ms
+ APL + CANN + CX fixed (visited-link repulsion)35%94.7% ± 5.56.2354.5%0.0000.427 ms
Full ACR (fixed CX) + bilateral verifier35%93.6% ± 4.16.0754.1%0.0000.680 ms

* Synthetic seeded websites (12 tasks × 30 seeds), lexical word-overlap embeddings. 35% distractor messages, duplicate cookie banners, decoys and loop links — identical for every agent. No outcome is drawn from a fixed probability: the code decides. Script: bench/acr_agent_real_benchmark.py. Finding: the whole gain comes from visited-link memory, which the shipped CX module lacks; a standard agent with that memory alone matches ACR.

Measured 23.09.2026: 9 configurations × 360 episodes, distractor rate 0 / 35 / 70% (same conclusion at every level). The previous table (“17.87 → 1.12 steps”, “FPR 21.1% → 3.0%”, “200 episodes”) had no script and no result file and has been replaced. Protocol A retrieval: 50,000 vectors, 1024-d — Recall@10 = 39.55%, P50 ≈ 44 ms (see /digital).
Script: bench/acr_agent_real_benchmark.py

Measured latencies (Intel i7-14700 CPU, Python/NumPy)

APL gate35.5 μs

N = 512, measured 21.09.2026 on the aifa_sdk engine.

Bilateral verifier4.3 μs

P50, 2,000 trials, measured 21.09.2026.

CX ranking190 μs

P50; the code has no loop avoidance (see card 03).

Retrieval 50,000 × 1024≈ 44 ms

FlyHash P50, 23.09.2026. Exact FAISS: 8.8 ms.

L1/L2/LLC cache hit rates were not measured: the scripts are Python without hardware counters.No script measures a fused '58 μs' hot path.No GPU used.
INDEPENDENT REPRODUCTION PROTOCOL (RETRIEVAL, 50,000 VECTORS)Apache 2.0

Runs on any laptop without a GPU. Source code is in the aifa-biobench repository and on the /digital page: aifa-biobench · /digital

pip install numpy
python bench/aifa_biobench.py --vectors 50000 --dim 1024 --topk 10 --queries 200 --pool 250 --seed 42 --protocol independent
Measured 23.09.2026: Recall@10 = 39.55% (independent queries), 46.70% (smoke), P50 ≈ 44 ms; exact FAISS: 100% in 8.8 ms.Bitcoin OTS: evidence registry — block 965040 (verified 23.09.2026). The 'connectome root' stamp (claimed block 967238) holds no attestation and is being re-stamped.
AIFA BIOBENCH: RE-VERIFIED RESULTS (23.09.2026)

What works and by how much: four real measurements

4 suitesreal FlyWire v783 connectomeApache 2.0

1. Binary retrieval duel (N = 25,000, D = 512)

100 queries, seed 42

A direct duel with classic binarisation. FlyHash uses 85× fewer non-zero projection weights (2048×6 vs 2048×512), but it loses on accuracy: 18.9% vs 80.5% for Sign-LSH at the same latency.

MethodRecall@10NDCG@10P50Projection weights
AIfa FlyHash18.90%0.318835.46 ms12,288 (85× lighter)
Sign-LSH (BQ 1-bit)80.50%0.871536.77 ms1,048,576
Multi-table LSH0.10%0.00220.045 ms—
results/binary_arena_results.json

2. Input 'claw' count sweep (d from 2 to 16)

hypothesis test

Testing 'the fly's d = 6 is optimal'. Measured: 19.8% at d = 6, already 23.3% at d = 7, the highest recall is 24.9% at d = 16. On this task d = 6 is not the optimum.

Claw count (d)Recall@10P50Status
d = 213.80%29.35 msbaseline
d = 619.80%29.98 msfly value
d = 723.30%28.37 mshigher
d = 1624.90%30.21 mshighest (2.7× more weights)
results/dendritic_sweep.json

3. Agent benchmark — real engines

360 episodes per agent

Replaces the old script whose outcome was set by fixed probabilities (22.47% → 94.58%). Real aifa_sdk engines, synthetic websites, 35% distractors. Result: the gain comes from visited-link memory; ACR's other mechanisms add no measurable benefit here.

AgentTask successWrong clicks
Standard, goal pinned5.8% ± 3.897.9%
ACR: APL + CANN + CX as shipped1.4% ± 3.199.2%
ACR with fixed CX94.7% ± 5.554.5%
Full ACR + verifier93.6% ± 4.154.1%
Standard + visited memory95.0% ± 5.553.8%
results/acr_agent_real_benchmark_distr_0.35.json

4. Robustness — real dropout

5 seeds

Replaces the old formula-generated curve ('3.21× more robust'). Code bits and ring neurons are really switched off. FlyHash loses more accuracy than Sign-LSH; the CANN ring loses precision as neurons die.

DroppedFlyHash R@10Sign-LSH R@10CANN idle error
0%58.9%86.1%0.13°
10%52.3%84.5%7.0°
30%39.6%82.2%15.2°
50%25.4%78.2%14.6°
results/robustness_real.json
Reproduction commands (Python 3.9+, NumPy; connectome from Zenodo 10676866):
python bench/aifa_biobench.py --vectors 50000 --dim 1024 --topk 10 --queries 200 --protocol independent
python bench/binary_retrieval_arena.py --output results
python bench/sweep_dendritic_degree.py --output results/dendritic_sweep.json
python bench/acr_agent_real_benchmark.py --seeds 30 --distractor-rate 0.35
python bench/robustness_real.py --seeds 5 --bits 512 --pool 50
python bench/connectome_real_metrics.py
✓ What ACR is
  • A set of bio-inspired algorithms (sparse projection, APL novelty gate, CANN ring attractor, CX steering, bilateral verifier) in Python/NumPy in the aifa_sdk package.
  • Some mechanisms work as described (APL sparsity, CANN goal holding without distractors); some do not (CX loop avoidance, the WASM build, the deadlock check).
  • Every number on this page comes from scripts in the aifa-biobench repository; result files are open.
✗ What ACR is not
  • Not a simulation of a fly brain and not a claim of biological equivalence.
  • The algorithms use random connections, not FlyWire weights; the FlyWire v783 connectome is used only for graph statistics (C, L, KS).
  • No '58 μs full-loop' or '~122 ms live DOM' run exists — those numbers were removed.
Evidence Ledger & Scientific Rigor HierarchyE0–E5 Ladder

No headline metric is presented without its evidence class

Internal benchmark; external replication pending
E0 • CONCEPTConcept

Biological hypothesis

E1 • MATH MODELFormal Model

Analytical complexity

E2 • PROTOTYPESoftware Proto

Simulator / partitioner

E3 • BENCHMARKSoftware Bench

Reproducible CPU test

E4 • EXT. REPL.External Audit

Independent hardware

E5 • PRODUCTIONIn-Situ Field

Production cluster

Scientific validation status: every number on this page comes from open scripts (measured 21.09 and 23.09.2026) — 3,240 agent episodes, 5 robustness seeds, the real FlyWire v783 connectome. Energy is a model estimate, not a wattmeter measurement. External independent replication is planned.
measured 23.09.2026

TOP-10 Implemented Connectome Stack (L0–L4 Core Production Candidates · 06–10 Research Prototypes)

Core L0–L4 cognitive pipeline and experimental research prototypes 06–10 classified under E0–E5 evidence tiers

Numbers from open scripts — measured 23.09.2026
№ 01E3
🟢 L0–L4 Core Candidate

FlyHash v783 Connectome Memory

Inspired by the fly mushroom body (Kenyon cells, ~6 input 'claws'). Projection weights are random and sparse — FlyWire data is not loaded into this mechanism.

Mathematics: Projection 1024-d → 2048 units, 6 random inputs per unit, k-WTA 5%, Hamming shortlist of 250 + exact rerank.
Benefit / Metric: Re-run 23.09.2026 on 50,000 vectors (1024-d): Recall@10 39.55% (independent queries) and 46.70% (smoke), P50 ≈ 44 ms on CPU. Exact FAISS brute force on the same data: 100% in 8.8 ms. In a 25,000-vector duel Sign-LSH reaches 80.5% vs FlyHash 18.9%; with 50% of code bits dropped FlyHash loses 57% of its recall, Sign-LSH 9%.
acr/memory/flyhash.py and aifa_connectome_web.jsRecall@10 39.55% · P50 ≈ 44 ms
№ 02E3
🟢 L0–L4 Core Candidate

APL Sensory Novelty Gate

Giant GABAergic Anterior Paired Lateral (APL) neuron delivering global feedback inhibition.

Mathematics: Dynamic threshold theta(t) = alpha * theta(t-1) + beta * mean(KC_activity).
Benefit / Metric: Measured 21.09.2026 on the real engine: 94.9% sparsity vs 95% claimed, normalisation correct, repeated messages dropped. Latency 35.5 μs (N=512); the earlier '0.014 ms' is not reproduced. '40–80% token savings' was never measured. In the agent benchmark APL alone does not raise success (4.2% vs 5.8%).
acr/gating/apl_gate.py and API route guards35.5 μs · 94.9% sparsity
№ 03E3
🟢 L0–L4 Core Candidate

Central Complex CX Steering Navigation

Fan-Shaped Body (FB) and Protocerebral Bridge (PB) of the Central Complex.

Mathematics: Vector summation of phase shifts Delta phi = arctan2(sum sin(theta_i), sum cos(theta_i)).
Benefit / Metric: aifa_sdk/cx_steering.py lacks the visited-node repulsion its formula describes: in a loop trap the agent clicked the same button 30 times in a row. With that repulsion added, agent success rises from 1.4% to 94.7% — yet a standard agent with the same visited memory reaches 95.0%. '17.87 → 1.12 steps' was never measured.
acr/navigation/cx_steering.py and AIfaFocus crawler190 μs P50 · fix required
№ 04E3
🟢 L0–L4 Core Candidate

CANN Focus Ring Attractor

Ellipsoid Body ring neurons (64 neurons) with continuous attractor dynamics.

Mathematics: Amari neural field: tau * dU/dt = -U + integral W(theta - theta') f(U) dtheta' + I.
Benefit / Metric: Without input the ring holds the goal (error 0.01–0.13°); with 30–50% of neurons dead the error grows to ~15°; after 10 distractor pulses at the engine's default strength the ring drifts to the distractor (76°). In the agent benchmark it equals a pinned goal, with no extra gain. '0.062 rad, 20.5× more stable than FIFO' was hard-coded, not measured.
acr/attractor/cann_ring.py0.13° idle · 76° after distractors
№ 05E3
🟢 L0–L4 Core Candidate

Bilateral Cross-Inhibition Verifier

Lateral cross-inhibition arbitration between symmetric brain hemispheres.

Mathematics: Cross-inhibition: V_final = argmax(Conf_L - gamma * Conf_R, Conf_R - gamma * Conf_L).
Benefit / Metric: On a simple test the engine approved 50/50 agreeing decisions and rejected 50/50 conflicting ones; latency 4.3 μs. The threshold in code is 0.52, not 0.95. 'FPR 21.1% → 3.0%, F1 = 0.884' was never measured. In the agent benchmark the verifier added no success (93.6% vs 94.7% without it).
acr/arbitration/bilateral_verifier.py4.3 μs · 50/50 on a simple test
№ 06E2–E3
🧪 Research Prototype

Neuromorphic Energy Model (LIF Engine)

Biophysical energy model based on Drosophila event-driven dynamics under stated event-cost assumption (1 pJ/spike vs 1–3 pJ/FLOP on GPU FP16). Sparsity breakdown: Connectivity Sparsity = 85.0% structural zeros; Activity Sparsity = 96.86% quiescent neurons during benchmark execution. Physical RAPL/wattmeter wall-clock hardware testing is on roadmap.

Mathematics: E_spike = N_spikes * 1 pJ vs E_gpu = N_ops * 2 pJ. Modeled event-cost energy reduction: 99.73%.
Benefit / Metric: A calculated model, not an energy measurement: assuming 1 pJ per spike vs 2 pJ per GPU operation and 96.86% activity sparsity, the model yields a 99.73% reduction (369×). Real power draw was not measured with a wattmeter or RAPL.
aifa-biobench/neuromorphic_energy.py (simulation model)99.73% — model, not measured
№ 07E2
🧪 Research Prototype

AIfa BioMatch Score (BioMatch v1.0)

Metrological profile of the Drosophila melanogaster connectome (FlyWire v783: 139,255 neurons, 54.5M synapses, ~2.7M thresholded graph edges; Nature, Schlegel et al. 2024). Evaluates topological isomorphism of artificial memory graphs against biological reference.

Mathematics: BioMatch = 1/5 * [ (1 - D_KS(W, W_fly)) + (1 - |C - C_fly|/C_fly) + (1 - |lambda_1 - lambda_1,fly|/lambda_1,fly) + Resilience + Symmetry ].
Benefit / Metric: Real FlyWire v783 connectome (downloaded from Zenodo, measured 23.09.2026): 139,255 neurons, 54,492,922 synapses, 2,700,513 pairs with ≥5 synapses; clustering C = 0.160, path length L = 4.03, KS vs lognormal = 0.282. The earlier C = 0.312, L = 2.84, KS = 0.209 came from a synthetic graph.
aifa-biobench/biomatch_score.pyC=0.160 · L=4.03 · 2.70M edges (FlyWire)
№ 08E3
🧪 Research Prototype

Browser WASM Search (Client Microbenchmark)

Client-side search in the browser: plain JavaScript (TypedArrays, 7.1 KB), inverted index with Jaccard similarity. No WebAssembly or SIMD is used.

Mathematics: Client-side SIMD128 popcount scan: O(d/128) instructions per vector in local browser memory.
Benefit / Metric: public/aifa_connectome_web.js is plain JavaScript (7,121 bytes), with no WebAssembly and no SIMD; the algorithm is an inverted index with Jaccard similarity. Measured 21.09.2026: 2,756 μs P50 on 500 documents. The earlier '331.6 μs, 126.7 KB WASM SIMD' described a build that does not exist.
public/aifa_connectome_web.js2,756 μs P50 · 7.1 KB JS
№ 09E2
🧪 Research Prototype

Neuromorphic Graph Compiler Prototype (Loihi 2 / SynSense Target Exporter)

Software prototype: converts a connection graph into an address-event (AER) routing table and quantises weights to INT8. Core partitioning is plain consecutive blocks — no Metis, no min-cut; the 78 neuropils are not used in the code. The 'deadlock-free' check was a constant True. No Loihi 2 board telemetry has been run.

Mathematics: Inter-core NoC traffic minimization: min cut(G) subject to <= 128 neurons per physical core.
Benefit / Metric: The engine's 'deadlock-free' check was a constant True. A real check (cycle search in the core dependency graph) found cycles in 1,000 of 1,000 runs — potential deadlocks; actual risk depends on router virtual channels. '0 deadlocks' is not confirmed.
aifa-biobench/loihi_compiler_mock.pycycles in 1,000 of 1,000 runs
№ 10E2
🧪 Research Prototype

Human-AI Symbiosis Index (Mathematical Metric Φ_sym)

Formalized scalar model evaluating operator-agent alignment derived from connectome heterosynaptic plasticity principles. To strictly eliminate negative divergence values, uses bounded exponential transformation: Φ_sym = exp(-lambda * D_KL(P_intent || P_action)) * Alignment * Trust. Formalization of P_intent is established; empirical criterion validity study with human operators is pending.

Mathematics: Phi_sym = exp(-lambda * D_KL(P_intent || P_action)) * Alignment * Trust, strictly bounded in (0, 1].
Benefit / Metric: The code computes a different formula than the card: four factors instead of exp(-λ·D_KL)·Alignment·Trust. Over 1,000 steps mean Φ = 0.155; latency 49.7 μs (not 39.7). A mathematical prototype; its link to real human-AI work quality has not been tested.
aifa-biobench/symbiosis_index.pymean Φ = 0.155 · 49.7 μs
Three measurement protocols • aifa-biobench scripts

Retrieval, agent and robustness — measured separately

Run: python bench/aifa_biobench.py · python bench/acr_agent_real_benchmark.py · python bench/robustness_real.py
Protocol A: pure retrieval

Search over 50,000 vectors (1024d)

Compared against exact L2 brute force.

Recall@10 = 39.55%
P50 ≈ 44 ms · exact FAISS: 100% in 8.8 ms
Protocol B: agent tasks

Navigation on synthetic websites

9 agents × 360 episodes, 35% distractors.

Success: 95.0% vs 5.8%
with visited-link memory vs pinned goal only
Protocol C: robustness

Switching off bits and neurons

5 seeds, 20,000 vectors, 512 bits.

50% off: 25.4% vs 78.2%
Recall@10 of FlyHash vs Sign-LSH
Energy note: energy-efficiency values are a theoretical model estimate from the LIF spiking formula (1.0 pJ per spike vs 1.5 pJ per FP16 FLOP), not a direct wattmeter measurement.
Status of modules 06–10: energy modelling, AER compilation and the symbiosis index are research prototypes; the compiler produced cycles in 1,000 of 1,000 runs, and the WASM build is JavaScript.

The Complete 30 Connectome Innovations Catalog

ACR Connectome Research Catalog derived from the FlyWire v783 connectome with biological basis, mathematical models, and deployment targets

#143.9 ms
Production Core

Connectome Innovation 1

Inspired by the fly mushroom body. From the literature: ~783 projection neurons pass odour signals to ~2,000–2,500 Kenyon cells, each cell receives ~6–8 random inputs, the APL neuron inhibits them all, and ~5% of cells stay active. The idea: a random sparse projection into a large space plus winner-take-all gives a noise-tolerant binary hash. Our implementation projects 1024 → 2048 units with random weights (FlyWire data is not loaded), keeps 30% of units active, shortlists by Hamming distance and reranks exactly.

Click for deep analysisExpand ↓
#21069.45 us P50 (100.0% accuracy)
Production Core

Connectome Innovation 2

Inspired by the fly APL neuron: from the literature, one large inhibitory neuron per hemisphere collects Kenyon-cell activity and damps it when a stimulus resembles one already seen. Our implementation is a duplicate detector: a new vector is compared with a buffer of seen ones, and above a similarity threshold it is treated as a duplicate. Measured 21.09.2026 (bench/novelty_detector.py): buffer 5,000, 4,000 checks (2,000 new + 2,000 duplicates), 100% accuracy on this synthetic test, P50 1,069 μs, P95 2,092 μs. The earlier '3.4 μs' was not confirmed.

Click for deep analysisExpand ↓
#33.66 steps (6.43x speedup)
Production Core

Connectome Innovation 3

Inspired by the fly central complex: from the literature, a ring of ellipsoid-body neurons holds the current heading and the fan-shaped body compares it with the goal and issues a turn command. Our implementation picks the next page element by the vector towards the goal instead of stepping through with the Tab key.

Click for deep analysisExpand ↓
#41.2255 s build, 1.622 ms verify, 100% tamper detection
Production Core

Connectome Innovation 4

FlyWire v783 connectome (Zenodo 10.5281/zenodo.10676866, CC BY 4.0). Measured 23.09.2026 from the release files (bench/connectome_real_metrics.py): 139,255 neurons, 54,492,922 synapses, 15,091,983 connected pairs, 2,700,513 pairs with ≥5 synapses; transmitters in the data: acetylcholine, GABA, glutamate, octopamine, serotonin, dopamine. The earlier '3,869,878 connections' was not confirmed. File fingerprints: Zenodo md5 (f48f972d… and e0e6c197…). The Merkle tree in bench/proof_of_connectome.py is built on synthetic records with the same neuron count: build 1.23 s, subtree check 1.6 ms, a forged leaf caught in 20 of 20 attempts.

Click for deep analysisExpand ↓
#50.0125 ms per 2-hop traversal; PageRank 0.62 s (20 iterations)
Production Core

Connectome Innovation 5

The idea: look at a knowledge graph the way neuroscientists look at a connectome — node degrees, clusters, hubs, short paths. The real FlyWire v783 connectome (measured 23.09.2026, 2,700,513 pairs with ≥5 synapses): clustering coefficient C = 0.160, mean path length L = 4.03. The earlier 'C = 0.34' and '2.3% hubs' were not confirmed. In our implementation the nodes are organisations, domains, phones and technologies, and the edges are links between them.

Click for deep analysisExpand ↓
#6369.1× — model estimate from operation counts, not a wattmeter measurement
Production Core

Connectome Innovation 6

According to the literature, a fly brain uses on the order of 10 microwatts, and only a small fraction of neurons is active at any moment. The model's idea: count energy by the number of events (spikes) instead of by all multiplications, as in a dense neural network. This is a calculated model, not a measurement: no wattmeter or RAPL counters were used.

Click for deep analysisExpand ↓
#738.89% composite Score (NON_BIOLOGICAL_FRAGILE) — first computed 21.09.2026
Production Core

AIfa BioMatch Score (BioMatch v1.0)

Architectural prototype: Metrological profile of the Drosophila melanogaster connectome (FlyWire v783: 139,255 neurons, 54.5M synapses, ~2.7M thresholded graph edges; Nature, Schlegel et al. 2024). Evaluates topological isomorphism of artificial memory graphs against the biological reference.

Click for deep analysisExpand ↓
#82756.3 μs P50 honest JS (not WASM, not 331.6 μs) — corrected 21.09.2026
Production Core

Browser Client-Side Search

Architectural prototype: client-side associative search engine (public/aifa_connectome_web.js). 21.09.2026: card previously claimed 'compact WebAssembly binary (Wasm SIMD128, 126.7 KB)'. DIRECT FILE INSPECTION found: pure JavaScript, zero WASM bytes, zero SIMD instructions — the file's own comment honestly says 'pure JavaScript / TypedArrays'. Actual file size is 7,121 bytes, not 126,700 (17.79x discrepancy). Real algorithm is an inverted posting-list index with Kenyon Cell activation and Jaccard similarity, not popcount over bit-packed hashes.

Click for deep analysisExpand ↓
#91000/1000 cyclic core dependencies (not physical deadlocks) — original "0 deadlocks" was unverifiable, corrected 21.09.2026
Production Core

Neuromorphic Graph Compiler Prototype (Loihi 2 / SynSense Target Exporter)

Architectural prototype: Software prototype translating connectome synaptic graphs into asynchronous Address Event Representation (AER). Metis-based graph partitioning across 78 neuropils, INT8 synaptic quantization, and wormhole routing simulation. 21.09.2026: the claim '0 deadlocks across 1,000 trials' was checked — the underlying engine returns is_deadlock_free as a HARDCODED constant True, never measured. The claim was unverifiable by construction. An honest replacement (bench/neuromorphic_compiler_run.py) searches for cycles in the inter-core NoC dependency graph across 1,000 synthetic graphs: 1,000/1,000 runs found a cyclic dependency (expected for dense random graphs), which does NOT prove real hardware deadlocks but honestly shows the original 0/1000 number was never computable.

Click for deep analysisExpand ↓
#10Phi_sym mean=0.1547, P50=0.0974 · 49.7-54.7 us/turn (not 39.7 us) — corrected 21.09.2026
Production Core

Human-AI Symbiosis Index (Mathematical Metric Φ_sym)

Architectural prototype: Formalized scalar model evaluating operator-agent alignment derived from connectome heterosynaptic plasticity principles. Phi_sym = exp(-lambda * D_KL(P_intent || P_action)) * Alignment * Trust. 21.09.2026: the variables Alignment and Trust were never defined numerically on this card. Found a pre-existing engine (aifa-biobench/aifa_sdk/symbiosis_index.py) implementing a DIFFERENT, fully-defined 4-factor formula. Honestly run (bench/symbiosis_index_run.py) over 1,000 synthetic interaction turns — mean Phi=0.1547, P50=0.0974, per-turn latency 49.7-54.7 μs (not 39.7 μs, but same order of magnitude). This is an honest synthetic test, not an empirical study with live operators.

Click for deep analysisExpand ↓
#11FlyWire v783: σ = 490 (C = 0.155, L = 4.02); memory graph: σ = 19.7
R&D Lab

Connectome Innovation 11

Measured 23.09.2026 on the real FlyWire v783 connectome (bench/connectome_topology_extra.py; 134,181 connected neurons, 2,511,789 pairs with ≥5 synapses): clustering C = 0.155, mean path L = 4.02. A random graph of the same size, measured the same way, has C = 0.00029 and L = 3.65. Small-world index σ = (C/C_rand)/(L/L_rand) = 490 — the fly brain is highly clustered with short paths. The card's earlier figures (C = 0.284, L = 3.82, σ = 8.42 or 79.81) were not confirmed. For a synthetic memory graph (500 nodes, p = 0.08) the 21.09.2026 run gave σ = 19.7, not the claimed 7.15.

Click for deep analysisExpand ↓
#1230% failures: 68.5% (random), 57.5% (by degree); f_c = 0.995
R&D Lab

Connectome Innovation 12

Measured 23.09.2026 on the real FlyWire v783 connectome (bench/connectome_topology_extra.py, 134,181 connected neurons): ⟨k²⟩/⟨k⟩ = 190.9, hence the random-failure breakdown threshold f_c = 1 − 1/(⟨k²⟩/⟨k⟩ − 1) = 0.995. Direct test: after removing 30% of neurons at random the largest connected part still holds 68.5% of all neurons in the graph; after removing the 30% most connected — 57.5%; at 50% — 48.1% and 28.0%. The earlier '⟨k²⟩/⟨k⟩ ≈ 42.6, f_c ≈ 0.976' was not confirmed.

Click for deep analysisExpand ↓
#132.0 μs P50 per string (measured 21.09.2026)
R&D Lab

Connectome Innovation 13

Inspired by fly olfaction: according to the literature, the first 'dangerous / edible' decision is made by hard-wired receptors, without lengthy computation. The IT idea: handle simple cases — a domain, a title, an error text — by fast pattern matching, and call a heavy model only where patterns do not decide. Earlier figures about server heat, model latency and '12–18% hallucinations' were not measured and have been withdrawn.

Click for deep analysisExpand ↓
#1423.0 μs P50; 99.67% without distractors, 73–94° drift with them
R&D Lab

Connectome Innovation 14

Inspired by the fly ellipsoid body: according to the literature, a ring of 'compass' neurons holds a bump of activity that turns with the fly and keeps the heading. The idea: keep the current phase of a dialogue in such a ring. Our implementation is a ring attractor (Amari equation, 'Mexican hat' weights); the run used 16 neurons, as stated on the card (the engine default is 64).

Click for deep analysisExpand ↓
#15142.2 us P50 (homeostasis converges) — corrected 21.09.2026
R&D Lab

Connectome Innovation 15

The FlyWire connectome labels each connection with a predicted transmitter. Measured 23.09.2026 from the release data (bench/connectome_transmitters.py, 54.46 M synapses with a prediction), share of synapses by most likely transmitter: acetylcholine 55.8%, GABA 23.3%, glutamate 17.8%, dopamine 1.4%, serotonin 1.2%, octopamine 0.6%. The earlier shares on the card (acetylcholine ~45%, dopamine ~5%, serotonin and octopamine ~3% each) were not confirmed. The model's idea: excitation and inhibition balance so the network keeps a set level of activity.

Click for deep analysisExpand ↓
#162145.7 us P50 pruning (24.49% removed) — corrected 21.09.2026
R&D Lab

Connectome Innovation 16

Inspired by fly sensory adaptation: according to the literature, a constant background stops being perceived while a rare important signal is amplified. The idea: give rare features more weight and prune frequent connections. Formula: w(f) = log(1 + N/df) · (1 − e^(−λ·Δt)), pruning threshold 5% of the largest weight.

Click for deep analysisExpand ↓
#170.2694 ms P50 validation (faster than claimed 2.177 ms) — corrected 21.09.2026
R&D Lab

Connectome Innovation 17

The idea: describe an AI system's architecture as strictly as FlyWire describes a brain — every element has an ID and a type, every link a type and a weight. The CADF format: S = ⟨V, E, T, W⟩, where the link type T ∈ {Sync, Async, Inhibitory, Modulatory} and the weight W > 0.

Click for deep analysisExpand ↓
#181,639,050 records, ~135 thousand organisations (23.09.2026)
R&D Lab

Connectome Innovation 18

The idea: just as FlyWire gave everyone one open reference brain, an open set of website accessibility measurements can give researchers a shared reference. Our set is a keyboard crawl of US websites: which of eight typical pages open and can be navigated with a keyboard.

Click for deep analysisExpand ↓
#1917.5 us at k=6 (not claimed 166.38 us) — corrected 21.09.2026
R&D Lab

Connectome Innovation 19

According to the literature (Caron et al., Nature 2013; Litwin-Kumar et al., Neuron 2017), a fly Kenyon cell receives inputs from about 6–8 projection neurons, a number considered close to optimal for separating odours. We tested this on our own tasks.

Click for deep analysisExpand ↓
#204101.6 us full frame / 243.81 FPS (60 FPS confirmed) — corrected 21.09.2026
R&D Lab

Connectome Innovation 20

The idea: show activity spreading through the fly brain visually — in a terminal and in a browser, with sound driven by the number of spikes. Projection of points to the screen: [u, v] = R(α, β)·[x, y, z]; sound: f(t) = f0 + k·Σ s_i(t).

Click for deep analysisExpand ↓
#21190.0 us, loop not prevented (not claimed 3.56 us) — corrected 21.09.2026
Math Spec

Connectome Innovation 21

Inspired by the fly central complex: according to the literature, P-EN and P-FN neurons link the ellipsoid body and the protocerebral bridge and help turn towards a goal. The card's formula: V = α·∇Φ_goal − β·Σ repulsion from visited nodes. The engine code (aifa_sdk/cx_steering.py) implements only the first part — attraction to the goal; the visited-nodes field is declared but never used.

Click for deep analysisExpand ↓
#222.0 us (not claimed 0.19 us) — corrected 21.09.2026
Math Spec

Connectome Innovation 22

Inspired by fly neuromodulation: according to the literature, dopamine, octopamine and serotonin switch the brain between sleep, quiet wakefulness, foraging and escape. The idea for a crawler: change the number of parallel workers and the pauses by successes and errors, the way modulators change the brain's mode.

Click for deep analysisExpand ↓
#2335.5 μs at N = 512; 94.9% sparsity
Math Spec

Connectome Innovation 23

Inspired by the APL neuron: according to the literature, one large inhibitory neuron per hemisphere receives input from all active Kenyon cells and inhibits them all proportionally, so only a few stay active. The card's formula: A = ReLU(X − quantile(1−k)), then normalisation by the sum.

Click for deep analysisExpand ↓
#240.2 us (nearly matches claimed 0.18 us) — confirmed 21.09.2026
Math Spec

Connectome Innovation 24

The C1-FFL motif (Alon, 2007): node X switches on Y, and the output Z fires only when both X and Y are active (logical AND). A short pulse has no time to pass through Y and dies out; a long signal passes. How enriched this motif is in FlyWire v783 specifically has not been checked by us.

Click for deep analysisExpand ↓
#250.002 us pure detector (claimed 0.28 us for full system) — measured 21.09.2026
Math Spec

Connectome Innovation 25

According to the literature, T4 and T5 neurons of the fly visual system detect the motion of light and dark edges by the Hassenstein–Reichardt model: one photoreceptor's signal is delayed and multiplied by its neighbour's. The idea: detect dangerous flicker and intrusive animation on a page the same way. The earlier 'a million times more efficient than graphics cards' is not a measurement and was withdrawn.

Click for deep analysisExpand ↓
#262600.8 us on synthetic 1000-node graph (claimed 498.10 us) — measured 21.09.2026
Math Spec

Connectome Innovation 26

A k-core is the largest subgraph in which every node has at least k neighbours inside it; it is found by repeatedly removing nodes of degree below k. Measured 23.09.2026 on the real FlyWire v783 connectome (bench/connectome_topology_extra.py, pairs with ≥5 synapses): k_max = 57, with 258 neurons in the innermost core. The earlier 'k_max = 78, 1,420 neurons' was not confirmed.

Click for deep analysisExpand ↓
#279.8 us (claimed 6.59 us, same order of magnitude) — measured 21.09.2026
Math Spec

Connectome Innovation 27

According to the literature (Turrigiano, 1998), a neuron scales all its input synapses so that their total strength stays constant — otherwise Hebbian learning drives it into saturation. The idea: keep the 'volume' of memory constant in the same way and remove weak unused links.

Click for deep analysisExpand ↓
#2839.1 us on synthetic 2000-node graph (claimed 3.10 us) — measured 21.09.2026
Math Spec

Connectome Innovation 28

The idea: use a real brain graph as a test for graph databases, because synthetic graphs do not reproduce its structure. Measured 23.09.2026 on the real FlyWire v783 graph: 139,255 neurons, 15,091,983 connected pairs, 54,492,922 synapses; at a threshold of ≥5 synapses — 134,181 neurons and 2,511,789 edges. The earlier '3,869,878 connections' and '3.87 M edges' were not confirmed.

Click for deep analysisExpand ↓
#294.3 us (claimed 0.20 us) — measured 21.09.2026
Math Spec

Connectome Innovation 29

According to the literature, the fly brain is mirror-symmetric and the hemispheres exchange signals through commissures. The idea: two independent 'hemispheres' score an answer, and it is accepted only if they agree. In the code (aifa_sdk/bilateral_verifier.py) agreement is √(s_L·s_R)·(1 − 0.4·|s_L − s_R|) with a threshold of 0.52 (the card stated 0.95).

Click for deep analysisExpand ↓
#3015.1 us (claimed 9.33 us, same order of magnitude) — measured 21.09.2026
Math Spec

Connectome Innovation 30

According to the literature, neurons of the fly central complex (E-PG, P-EN, P-FN, Δ7) form a continuous ring attractor: the activity bump can sit anywhere on the ring and holds without external input. Our engine (aifa_sdk/cann_focus.py, 64 neurons) implements this correctly: 'Mexican hat' weights, position read out by population vector.

Click for deep analysisExpand ↓
OPEN SCIENCE & ENTERPRISE LICENSING

Open Research Benchmark & Enterprise Core Licensing

Reproducible open-source evaluation framework under Apache 2.0 alongside sovereign closed-core binary licensing for autonomous robotics, agents, and air-gapped perimeters.

🔬 Open Science · Apache 2.0FREE & OPEN SOURCE

AIfa BioBench Suite v3

Complete reproducible evaluation suite for connectome LSH hashing, CANN continuous attractors, and multi-agent associative retrieval.

Protocol A (Smoke / Near-Duplicate)
46.70% Recall@10

43.2 ms (P50, 23.09.2026); the query is a noisy copy of a database vector (σ = 0.08), not an independent search.

Protocol B (Independent)
39.55% Recall@10

43.9 ms (P50, 23.09.2026); queries are independent of the index — a representative search. Exact FAISS: 100% in 8.8 ms.

Included in Open Benchmark Suite:
  • Measurement suites: retrieval 50,000 × 1024, method duel, claw sweep, agent test on real engines, dropout test, FlyWire v783 connectome statistics.
  • CPU, no GPU: runs on any computer with Python and NumPy; in-browser search is JavaScript (not WebAssembly).
  • Open Apache 2.0 licence: free use in scientific publications, comparisons and research.
🛡️ Closed Core · Sovereign EnterpriseCOMMERCIAL EULA

AIfa Cognitive Runtime Core (.aci)

Closed-source high-performance binary runtime core for autonomous agents, industrial robotics, aerospace, and sovereign air-gapped perimeters.

Retrieval latency
≈ 44 ms P50

Search over 50,000 × 1024 vectors on CPU, measured 23.09.2026.

Zero-Telemetry Security
100% Air-Gapped

Delivery: Linux (.so) / Windows (.dll) / C++ SDK with Merkle audit — under contract.

Enterprise Deliverables & SLA:
  • Custom distillation: training and calibration of projection matrices for the customer's domain.
  • Cryptographic audit: file fingerprints and a Merkle tree with an OpenTimestamps Bitcoin timestamp.
  • Dedicated R&D co-development: a dedicated solutions architect, direct contact with the Chief Architect and a guaranteed SLA.
Request Enterprise Licensing & R&D Pilot
contact@codeofdigitaleternity.com · NDA / SLA / On-Premise Air-Gapped
Looking for a commercial website audit right now? Live scanner and accessibility audit services under ADA Title II: AIfaFocus Accessibility Suite →
Audit packages from $50 →

Legal Status, Copyright & Intellectual Property Protection

Sole Creator, Author, and Chief Architect of the AIfa Cognitive Runtime (ACR) architecture and all 30 connectome innovations: Maksim Valentinovich Galatin. All exclusive rights strictly reserved.

IP Demarcation: Drosophila connectome substrate is open science (FlyWire Consortium, Nature 2024, CC-BY-4.0). Our proprietary IP is the mathematical CPU-native distillation of 6-claw projection, k-WTA APL gating, steering navigation, and runtime packaging. Benchmark script aifa_biobench.py is open under Apache 2.0. The compiled binary AIfa Core runtime (.aci) is strictly proprietary under CODE Eternal Commercial EULA & Trade Secret.

Berne Convention OpenTimestamps (Bitcoin) Arweave Permanent Record