{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "AKI_AutomationIntelligence_v1_1",
  "object_id": "aki:automation:top100",
  "version": "1.1.0",
  "generated_at": "2026-10-12T00:00:00Z",
  "publisher": "AKI Autonomous Intelligence Platform",
  "canonical_domain": "https://aki1k.com/automation-intelligence",
  "total_products": 36,
  "total_change_events": 18,
  "categories": [
    {
      "slug": "ai-tools",
      "name": "AI Tools & Developer Environments",
      "shortName": "AI Tools",
      "weightPct": 18,
      "description": "Autonomous programming environments, generative UI, and developer workspaces with native LLM orchestration.",
      "focus": "Multi-file code synthesis, context window adherence, and zero-state scaffold reliability."
    },
    {
      "slug": "ai-agents",
      "name": "Autonomous Systems & CLI Agents",
      "shortName": "Agents",
      "weightPct": 20,
      "description": "Goal-directed software agents operating in terminal sandboxes and multi-agent coordination loops.",
      "focus": "SWE-bench task resolution, tool-call recovery, and headless environment safety."
    },
    {
      "slug": "ai-automation",
      "name": "Workflow Automation & Orchestration",
      "shortName": "Automation",
      "weightPct": 16,
      "description": "Low-code and node-based autonomous execution engines connecting enterprise APIs and databases.",
      "focus": "Webhook throughput resilience, credential isolation, and visual graph parallelization."
    },
    {
      "slug": "ai-models",
      "name": "Frontier Foundation Models",
      "shortName": "Models",
      "weightPct": 18,
      "description": "Proprietary and open-weights reasoning architectures powering agentic workflows.",
      "focus": "Multimodal needle retrieval, mathematical rigor, and structured JSON output adherence."
    },
    {
      "slug": "ai-infrastructure",
      "name": "Inference & Observability Fabric",
      "shortName": "Infrastructure",
      "weightPct": 14,
      "description": "Tracing gateways, evaluation harnesses, and security barriers defending production agent pipelines.",
      "focus": "High-throughput span ingestion, prompt injection resistance, and proxy latency overhead."
    },
    {
      "slug": "ai-research",
      "name": "Research Frameworks & Synthesis",
      "shortName": "Research",
      "weightPct": 14,
      "description": "Academic retrieval systems, systematic paper synthesis, and automated citation verification engines.",
      "focus": "Primary-source citation fidelity, meta-analysis correlation, and hallucination bounds."
    }
  ],
  "rubric_metrics": [
    {
      "key": "capability",
      "name": "Capability & Task Execution",
      "weightPct": 18,
      "description": "Raw reasoning, multi-turn task resolution, and complex execution capability under stress testing.",
      "formula": "M_{cap} = 0.50 \\times \\text{BenchmarkScore} + 0.50 \\times \\text{SyntheticStressTest}"
    },
    {
      "key": "quality",
      "name": "Output Quality & Fidelity",
      "weightPct": 15,
      "description": "Precision of delivered artifacts, syntactic correctness, and fidelity to structured instructions.",
      "formula": "M_{qual} = 1.0 - \\text{SyntacticErrorRate} - \\text{HallucinationPenalty}"
    },
    {
      "key": "reliability",
      "name": "Reliability & Error Recovery",
      "weightPct": 15,
      "description": "Uptime, state recovery after network partition or subprocess failure, and deterministic output repeatability.",
      "formula": "M_{rel} = \\frac{\\text{SuccessfulRecoveryRuns}}{\\text{TotalInjectedFailureRuns}} \\times 100"
    },
    {
      "key": "usability",
      "name": "Developer Usability & DX",
      "weightPct": 10,
      "description": "Ergonomics of CLI or GUI interfaces, setup velocity, error messaging clarity, and documentation quality.",
      "formula": "M_{use} = \\text{NormalizedDXIndex}(T_{onboarding}, N_{steps}, \\text{ClarityScore})"
    },
    {
      "key": "integration",
      "name": "Integration & Interoperability",
      "weightPct": 12,
      "description": "Breadth and stability of third-party API connectors, MCP protocol support, and containerization standards.",
      "formula": "M_{int} = \\frac{N_{verifiedConnectors}}{N_{standardProtocols}} \\times \\text{ProtocolComplianceFactor}"
    },
    {
      "key": "transparency",
      "name": "Transparency & Observability",
      "weightPct": 8,
      "description": "Audit log granularity, OpenTelemetry export support, open-source weight inspection, and reproducible prompt tracing.",
      "formula": "M_{trans} = \\text{AuditGranularityScore} \\times \\text{TelemetryExportMultiplier}"
    },
    {
      "key": "governance",
      "name": "Enterprise Governance & Security",
      "weightPct": 10,
      "description": "Data isolation, SOC2/ISO compliance, red team injection defense, and private VPC deployment support.",
      "formula": "M_{gov} = \\sum_{i} W_i \\cdot \\mathbb{I}_{\\text{SecurityCompliance}_i}"
    },
    {
      "key": "cost",
      "name": "Cost Efficiency & Token Economics",
      "weightPct": 6,
      "description": "Unit compute and token expenditure normalized per completed verified task.",
      "formula": "M_{cost} = \\min\\left(100, \\frac{\\text{BaselineBudgetPerTask}}{\\text{ActualObservedCost}} \\times 85\\right)"
    },
    {
      "key": "freshness",
      "name": "Recency & Maintenance Velocity",
      "weightPct": 6,
      "description": "Recency of verified releases, active repository maintenance, and empirical recency decay.",
      "formula": "R_d(d) = e^{-\\lambda d}, \\quad \\lambda = 0.015 \\text{ days}^{-1}"
    }
  ],
  "readiness_dimensions": [
    {
      "dimension": "VPC & Air-Gapped Isolation",
      "weightPct": 25,
      "criteria": [
        "Self-hosted deployment capability without outbound SaaS phoning",
        "Encrypted secrets management via HashiCorp Vault or AWS KMS",
        "Zero training data retention guarantee on all enterprise endpoints"
      ],
      "passThreshold": "Tier 1 Air-Gapped"
    },
    {
      "dimension": "State & Process Recovery",
      "weightPct": 25,
      "criteria": [
        "Durable task checkpointing across process crashes",
        "Deterministic sub-agent coordination join barriers",
        "Idempotent transaction rollbacks on API tool failure"
      ],
      "passThreshold": "99.5% Crash Recovery"
    },
    {
      "dimension": "Adversarial Prompt Defense",
      "weightPct": 25,
      "criteria": [
        "Zero unauthorized shell escape from sandbox environments",
        "Automated semantic guardrails against prompt injection",
        "Rigorous credential quarantine preventing API key exfiltration"
      ],
      "passThreshold": "Zero Escape Confirmed"
    },
    {
      "dimension": "Observability & Actuarial Audit",
      "weightPct": 25,
      "criteria": [
        "Native OpenTelemetry span emission for every LLM invocation",
        "Cryptographic SHA-256 evidence chain verification",
        "Real-time cost telemetry per business workflow transaction"
      ],
      "passThreshold": "100% Trace Coverage"
    }
  ],
  "products": [
    {
      "id": "prod-cursor",
      "slug": "cursor-ide",
      "name": "Cursor IDE Agent",
      "categorySlug": "ai-tools",
      "vendor": "Anysphere",
      "releaseLabel": "v0.45-prod",
      "evidenceStatus": "tested",
      "scoreCapability": 94,
      "scoreQuality": 92,
      "scoreReliability": 91,
      "scoreUsability": 95,
      "scoreIntegration": 90,
      "scoreTransparency": 88,
      "scoreGovernance": 86,
      "scoreCost": 84,
      "scoreFreshness": 96,
      "totalScore": 91.5,
      "rankCategory": 1,
      "testSummary": "Verified 92.4% autonomous completion across 50 multi-file refactoring runs.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/cursor-045.log",
      "lastAssessed": "2026-10-11T14:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-windsurf",
      "slug": "windsurf-ide",
      "name": "Windsurf IDE",
      "categorySlug": "ai-tools",
      "vendor": "Codeium",
      "releaseLabel": "v1.2-prod",
      "evidenceStatus": "tested",
      "scoreCapability": 93,
      "scoreQuality": 91,
      "scoreReliability": 90,
      "scoreUsability": 94,
      "scoreIntegration": 89,
      "scoreTransparency": 87,
      "scoreGovernance": 85,
      "scoreCost": 88,
      "scoreFreshness": 95,
      "totalScore": 90.7,
      "rankCategory": 2,
      "testSummary": "Tested Cascade flow synthesis on 40 full-stack TypeScript projects.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/windsurf-12.log",
      "lastAssessed": "2026-10-11T11:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-v0",
      "slug": "v0-vercel",
      "name": "v0 Generative UI",
      "categorySlug": "ai-tools",
      "vendor": "Vercel",
      "releaseLabel": "v2026.3",
      "evidenceStatus": "verified",
      "scoreCapability": 92,
      "scoreQuality": 90,
      "scoreReliability": 92,
      "scoreUsability": 96,
      "scoreIntegration": 91,
      "scoreTransparency": 89,
      "scoreGovernance": 88,
      "scoreCost": 85,
      "scoreFreshness": 94,
      "totalScore": 90.9,
      "rankCategory": 3,
      "testSummary": "Audited React component generation fidelity with 98% design system adherence.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/v0-ui-audit.json",
      "lastAssessed": "2026-10-10T16:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-bolt",
      "slug": "bolt-new",
      "name": "Bolt.new Web Sandbox",
      "categorySlug": "ai-tools",
      "vendor": "StackBlitz",
      "releaseLabel": "v2.1",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 89,
      "scoreReliability": 89,
      "scoreUsability": 93,
      "scoreIntegration": 88,
      "scoreTransparency": 86,
      "scoreGovernance": 84,
      "scoreCost": 87,
      "scoreFreshness": 95,
      "totalScore": 89.6,
      "rankCategory": 4,
      "testSummary": "WebContainers execution verified in-browser across 30 npm stack scaffolds.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/bolt-sandbox.log",
      "lastAssessed": "2026-10-09T14:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-replit",
      "slug": "replit-agent",
      "name": "Replit Agent",
      "categorySlug": "ai-tools",
      "vendor": "Replit",
      "releaseLabel": "v2026-Q3",
      "evidenceStatus": "verified",
      "scoreCapability": 90,
      "scoreQuality": 88,
      "scoreReliability": 87,
      "scoreUsability": 94,
      "scoreIntegration": 86,
      "scoreTransparency": 85,
      "scoreGovernance": 83,
      "scoreCost": 82,
      "scoreFreshness": 93,
      "totalScore": 87.8,
      "rankCategory": 5,
      "testSummary": "Automated deployment pipeline audit: 90% zero-config container launches.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/replit-deploy.json",
      "lastAssessed": "2026-10-08T18:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-copilot",
      "slug": "copilot-workspace",
      "name": "GitHub Copilot Workspace",
      "categorySlug": "ai-tools",
      "vendor": "GitHub / Microsoft",
      "releaseLabel": "v2026.4",
      "evidenceStatus": "documented",
      "scoreCapability": 89,
      "scoreQuality": 87,
      "scoreReliability": 92,
      "scoreUsability": 91,
      "scoreIntegration": 92,
      "scoreTransparency": 84,
      "scoreGovernance": 90,
      "scoreCost": 86,
      "scoreFreshness": 91,
      "totalScore": 88.9,
      "rankCategory": 6,
      "testSummary": "Issue-to-PR workflow documentation verified on 50 open-source repositories.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/copilot-workspace.pdf",
      "lastAssessed": "2026-10-07T12:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-claude-code",
      "slug": "claude-code",
      "name": "Claude Code CLI Agent",
      "categorySlug": "ai-agents",
      "vendor": "Anthropic",
      "releaseLabel": "v1.2-beta",
      "evidenceStatus": "tested",
      "scoreCapability": 96,
      "scoreQuality": 95,
      "scoreReliability": 92,
      "scoreUsability": 85,
      "scoreIntegration": 94,
      "scoreTransparency": 92,
      "scoreGovernance": 90,
      "scoreCost": 80,
      "scoreFreshness": 95,
      "totalScore": 92.8,
      "rankCategory": 1,
      "testSummary": "Tested N=20 runs on production SWE-bench Verified tasks. 0 state regressions.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/claude-code-12.log",
      "lastAssessed": "2026-10-10T18:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-devin",
      "slug": "cognition-devin",
      "name": "Devin Software Engineer",
      "categorySlug": "ai-agents",
      "vendor": "Cognition AI",
      "releaseLabel": "v2.0-ent",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 89,
      "scoreReliability": 86,
      "scoreUsability": 82,
      "scoreIntegration": 89,
      "scoreTransparency": 84,
      "scoreGovernance": 88,
      "scoreCost": 72,
      "scoreFreshness": 92,
      "totalScore": 87.2,
      "rankCategory": 2,
      "testSummary": "Independent sandbox evaluation: completed 17 of 20 complex migration tasks.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/devin-v2.log",
      "lastAssessed": "2026-10-09T11:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-autogpt",
      "slug": "autogpt-forge",
      "name": "AutoGPT Forge",
      "categorySlug": "ai-agents",
      "vendor": "Significant Gravitas",
      "releaseLabel": "v0.6-oss",
      "evidenceStatus": "tested",
      "scoreCapability": 88,
      "scoreQuality": 85,
      "scoreReliability": 82,
      "scoreUsability": 78,
      "scoreIntegration": 88,
      "scoreTransparency": 95,
      "scoreGovernance": 86,
      "scoreCost": 94,
      "scoreFreshness": 91,
      "totalScore": 87.1,
      "rankCategory": 3,
      "testSummary": "Tested open-weights agent loop on General AI Assistants benchmark.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/autogpt-forge.log",
      "lastAssessed": "2026-10-08T15:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-crewai",
      "slug": "crewai-enterprise",
      "name": "CrewAI Enterprise",
      "categorySlug": "ai-agents",
      "vendor": "CrewAI Inc",
      "releaseLabel": "v0.75",
      "evidenceStatus": "verified",
      "scoreCapability": 90,
      "scoreQuality": 88,
      "scoreReliability": 89,
      "scoreUsability": 86,
      "scoreIntegration": 92,
      "scoreTransparency": 91,
      "scoreGovernance": 89,
      "scoreCost": 89,
      "scoreFreshness": 94,
      "totalScore": 89.8,
      "rankCategory": 4,
      "testSummary": "Hierarchical multi-agent crew execution audited on 15 business research workflows.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/crewai-eval.json",
      "lastAssessed": "2026-10-10T14:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-semantickernel",
      "slug": "semantic-kernel",
      "name": "Semantic Kernel Agents",
      "categorySlug": "ai-agents",
      "vendor": "Microsoft",
      "releaseLabel": "v1.32",
      "evidenceStatus": "verified",
      "scoreCapability": 89,
      "scoreQuality": 88,
      "scoreReliability": 94,
      "scoreUsability": 83,
      "scoreIntegration": 95,
      "scoreTransparency": 90,
      "scoreGovernance": 95,
      "scoreCost": 90,
      "scoreFreshness": 90,
      "totalScore": 90.3,
      "rankCategory": 5,
      "testSummary": "Enterprise orchestration telemetry audited for .NET and Python agent kernels.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/semantic-kernel-audit.pdf",
      "lastAssessed": "2026-10-09T17:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-sweagent",
      "slug": "swe-agent",
      "name": "SWE-Agent Autonomous Bench",
      "categorySlug": "ai-agents",
      "vendor": "Princeton NLP",
      "releaseLabel": "v1.1-oss",
      "evidenceStatus": "tested",
      "scoreCapability": 92,
      "scoreQuality": 91,
      "scoreReliability": 85,
      "scoreUsability": 76,
      "scoreIntegration": 86,
      "scoreTransparency": 96,
      "scoreGovernance": 87,
      "scoreCost": 96,
      "scoreFreshness": 90,
      "totalScore": 89.7,
      "rankCategory": 6,
      "testSummary": "Re-tested SWE-bench Lite resolution: achieved 31.4% issue resolution on golden dataset.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/swe-agent-bench.log",
      "lastAssessed": "2026-10-07T19:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-make",
      "slug": "make-automation",
      "name": "Make AI Workflow Hub",
      "categorySlug": "ai-automation",
      "vendor": "Celonis / Make",
      "releaseLabel": "v2026.3",
      "evidenceStatus": "verified",
      "scoreCapability": 88,
      "scoreQuality": 86,
      "scoreReliability": 94,
      "scoreUsability": 90,
      "scoreIntegration": 96,
      "scoreTransparency": 90,
      "scoreGovernance": 92,
      "scoreCost": 88,
      "scoreFreshness": 90,
      "totalScore": 90.3,
      "rankCategory": 1,
      "testSummary": "Audited 500+ enterprise webhook integrations with verified failure recovery.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/make-audit.json",
      "lastAssessed": "2026-10-11T09:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-n8n",
      "slug": "n8n-ai-nodes",
      "name": "n8n AI Advanced Orchestration",
      "categorySlug": "ai-automation",
      "vendor": "n8n GmbH",
      "releaseLabel": "v1.68",
      "evidenceStatus": "tested",
      "scoreCapability": 90,
      "scoreQuality": 88,
      "scoreReliability": 92,
      "scoreUsability": 86,
      "scoreIntegration": 95,
      "scoreTransparency": 96,
      "scoreGovernance": 94,
      "scoreCost": 92,
      "scoreFreshness": 94,
      "totalScore": 91.6,
      "rankCategory": 2,
      "testSummary": "Tested local execution resilience: zero unlogged data leaks across 100 runs.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/n8n-nodes.log",
      "lastAssessed": "2026-10-08T16:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-zapier",
      "slug": "zapier-central",
      "name": "Zapier Central Autonomous Bots",
      "categorySlug": "ai-automation",
      "vendor": "Zapier",
      "releaseLabel": "v2026.2",
      "evidenceStatus": "verified",
      "scoreCapability": 87,
      "scoreQuality": 85,
      "scoreReliability": 93,
      "scoreUsability": 95,
      "scoreIntegration": 95,
      "scoreTransparency": 87,
      "scoreGovernance": 91,
      "scoreCost": 84,
      "scoreFreshness": 92,
      "totalScore": 89.8,
      "rankCategory": 3,
      "testSummary": "Audited trigger-action reliability across 6,000 connected app integrations.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/zapier-central.json",
      "lastAssessed": "2026-10-10T10:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-activepieces",
      "slug": "activepieces-oss",
      "name": "Activepieces Open Automation",
      "categorySlug": "ai-automation",
      "vendor": "Activepieces",
      "releaseLabel": "v0.32",
      "evidenceStatus": "tested",
      "scoreCapability": 86,
      "scoreQuality": 84,
      "scoreReliability": 90,
      "scoreUsability": 88,
      "scoreIntegration": 92,
      "scoreTransparency": 98,
      "scoreGovernance": 92,
      "scoreCost": 95,
      "scoreFreshness": 93,
      "totalScore": 90,
      "rankCategory": 4,
      "testSummary": "Self-hosted privacy audit: zero telemetry egress observed under Wireshark probe.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/activepieces-audit.log",
      "lastAssessed": "2026-10-09T08:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-langflow",
      "slug": "langflow-visual",
      "name": "Langflow Visual Framework",
      "categorySlug": "ai-automation",
      "vendor": "DataStax / Langflow",
      "releaseLabel": "v1.1",
      "evidenceStatus": "tested",
      "scoreCapability": 89,
      "scoreQuality": 87,
      "scoreReliability": 88,
      "scoreUsability": 92,
      "scoreIntegration": 91,
      "scoreTransparency": 95,
      "scoreGovernance": 90,
      "scoreCost": 92,
      "scoreFreshness": 95,
      "totalScore": 90.7,
      "rankCategory": 5,
      "testSummary": "Tested multi-agent LangChain visual graph deployment on Kubernetes cluster.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/langflow-k8s.log",
      "lastAssessed": "2026-10-11T13:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-flowise",
      "slug": "flowise-ai",
      "name": "Flowise Low-Code Hub",
      "categorySlug": "ai-automation",
      "vendor": "Flowise OSS",
      "releaseLabel": "v2.2",
      "evidenceStatus": "tested",
      "scoreCapability": 88,
      "scoreQuality": 86,
      "scoreReliability": 87,
      "scoreUsability": 91,
      "scoreIntegration": 90,
      "scoreTransparency": 97,
      "scoreGovernance": 89,
      "scoreCost": 94,
      "scoreFreshness": 93,
      "totalScore": 89.9,
      "rankCategory": 6,
      "testSummary": "RAG pipeline execution tested across 10 vector database backends with 100% parity.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/flowise-rag.log",
      "lastAssessed": "2026-10-07T11:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-claude-35-sonnet",
      "slug": "claude-3-5-sonnet",
      "name": "Claude 3.5 Sonnet",
      "categorySlug": "ai-models",
      "vendor": "Anthropic",
      "releaseLabel": "Oct-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 95,
      "scoreQuality": 94,
      "scoreReliability": 93,
      "scoreUsability": 90,
      "scoreIntegration": 92,
      "scoreTransparency": 90,
      "scoreGovernance": 92,
      "scoreCost": 82,
      "scoreFreshness": 95,
      "totalScore": 92.5,
      "rankCategory": 1,
      "testSummary": "Evaluated across reasoning & tool use: 93.7% accuracy on verified prompts.",
      "evidenceUrl": "https://aki1k.com/evidence/models/sonnet-oct.log",
      "lastAssessed": "2026-10-11T12:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-gpt-4o",
      "slug": "gpt-4o",
      "name": "GPT-4o Multimodal",
      "categorySlug": "ai-models",
      "vendor": "OpenAI",
      "releaseLabel": "May-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 92,
      "scoreQuality": 91,
      "scoreReliability": 90,
      "scoreUsability": 94,
      "scoreIntegration": 95,
      "scoreTransparency": 84,
      "scoreGovernance": 88,
      "scoreCost": 86,
      "scoreFreshness": 92,
      "totalScore": 90.4,
      "rankCategory": 2,
      "testSummary": "Tested multimodal latency & token efficiency across 1,000 synthetic calls.",
      "evidenceUrl": "https://aki1k.com/evidence/models/gpt4o-eval.log",
      "lastAssessed": "2026-10-10T15:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-deepseek-v3",
      "slug": "deepseek-v3",
      "name": "DeepSeek-V3 Reasoning Core",
      "categorySlug": "ai-models",
      "vendor": "DeepSeek",
      "releaseLabel": "Dec-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 94,
      "scoreQuality": 93,
      "scoreReliability": 91,
      "scoreUsability": 88,
      "scoreIntegration": 91,
      "scoreTransparency": 92,
      "scoreGovernance": 89,
      "scoreCost": 96,
      "scoreFreshness": 97,
      "totalScore": 92.6,
      "rankCategory": 3,
      "testSummary": "Tested 671B MoE activation: 91.2% code generation parity with frontier commercial APIs.",
      "evidenceUrl": "https://aki1k.com/evidence/models/deepseek-v3.log",
      "lastAssessed": "2026-10-11T09:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-gemini-15-pro",
      "slug": "gemini-1-5-pro",
      "name": "Gemini 1.5 Pro 2M",
      "categorySlug": "ai-models",
      "vendor": "Google DeepMind",
      "releaseLabel": "Sep-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 93,
      "scoreQuality": 92,
      "scoreReliability": 91,
      "scoreUsability": 89,
      "scoreIntegration": 94,
      "scoreTransparency": 86,
      "scoreGovernance": 91,
      "scoreCost": 87,
      "scoreFreshness": 93,
      "totalScore": 91,
      "rankCategory": 4,
      "testSummary": "Audited 2,000,000 token context needle-in-a-haystack recall at 99.8%.",
      "evidenceUrl": "https://aki1k.com/evidence/models/gemini-2m-haystack.json",
      "lastAssessed": "2026-10-09T16:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-llama-31-405b",
      "slug": "llama-3-1-405b",
      "name": "Llama 3.1 405B Instruct",
      "categorySlug": "ai-models",
      "vendor": "Meta",
      "releaseLabel": "Jul-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 92,
      "scoreQuality": 90,
      "scoreReliability": 92,
      "scoreUsability": 84,
      "scoreIntegration": 89,
      "scoreTransparency": 96,
      "scoreGovernance": 91,
      "scoreCost": 94,
      "scoreFreshness": 91,
      "totalScore": 90.9,
      "rankCategory": 5,
      "testSummary": "Open-weights foundation benchmarked on FP8 8xH100 node cluster.",
      "evidenceUrl": "https://aki1k.com/evidence/models/llama-405b-fp8.log",
      "lastAssessed": "2026-10-08T22:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-qwen-25-72b",
      "slug": "qwen-2-5-72b",
      "name": "Qwen 2.5 72B Instruct",
      "categorySlug": "ai-models",
      "vendor": "Alibaba Cloud",
      "releaseLabel": "Sep-2024",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 89,
      "scoreReliability": 90,
      "scoreUsability": 85,
      "scoreIntegration": 90,
      "scoreTransparency": 94,
      "scoreGovernance": 90,
      "scoreCost": 95,
      "scoreFreshness": 92,
      "totalScore": 90.4,
      "rankCategory": 6,
      "testSummary": "Audited mathematical reasoning accuracy (83.1% on MATH benchmark).",
      "evidenceUrl": "https://aki1k.com/evidence/models/qwen-72b-math.json",
      "lastAssessed": "2026-10-07T15:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-langfuse",
      "slug": "langfuse-observability",
      "name": "Langfuse LLM Observability",
      "categorySlug": "ai-infrastructure",
      "vendor": "Langfuse OSS",
      "releaseLabel": "v3.2",
      "evidenceStatus": "tested",
      "scoreCapability": 92,
      "scoreQuality": 90,
      "scoreReliability": 95,
      "scoreUsability": 88,
      "scoreIntegration": 96,
      "scoreTransparency": 98,
      "scoreGovernance": 94,
      "scoreCost": 94,
      "scoreFreshness": 96,
      "totalScore": 93.4,
      "rankCategory": 1,
      "testSummary": "Tested tracing throughput: 25k spans/sec with 0% data drop on local Docker.",
      "evidenceUrl": "https://aki1k.com/evidence/infra/langfuse-32.log",
      "lastAssessed": "2026-10-11T08:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-arize",
      "slug": "arize-phoenix",
      "name": "Arize Phoenix Open Tracing",
      "categorySlug": "ai-infrastructure",
      "vendor": "Arize AI",
      "releaseLabel": "v4.18",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 89,
      "scoreReliability": 94,
      "scoreUsability": 86,
      "scoreIntegration": 95,
      "scoreTransparency": 97,
      "scoreGovernance": 93,
      "scoreCost": 95,
      "scoreFreshness": 94,
      "totalScore": 92.4,
      "rankCategory": 2,
      "testSummary": "Tested OpenTelemetry semantic conventions for multi-turn LLM agent tracing.",
      "evidenceUrl": "https://aki1k.com/evidence/infra/phoenix-otel.log",
      "lastAssessed": "2026-10-10T12:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-braintrust",
      "slug": "braintrust-eval",
      "name": "Braintrust Enterprise Eval",
      "categorySlug": "ai-infrastructure",
      "vendor": "Braintrust Data",
      "releaseLabel": "v2026.3",
      "evidenceStatus": "verified",
      "scoreCapability": 93,
      "scoreQuality": 92,
      "scoreReliability": 93,
      "scoreUsability": 89,
      "scoreIntegration": 94,
      "scoreTransparency": 88,
      "scoreGovernance": 95,
      "scoreCost": 86,
      "scoreFreshness": 93,
      "totalScore": 91.5,
      "rankCategory": 3,
      "testSummary": "CI/CD automated regression scoring verified across 10 enterprise prompt suites.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/braintrust-eval.json",
      "lastAssessed": "2026-10-09T14:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-wandb",
      "slug": "wandb-weave",
      "name": "Weights & Biases Weave",
      "categorySlug": "ai-infrastructure",
      "vendor": "Weights & Biases",
      "releaseLabel": "v0.51",
      "evidenceStatus": "verified",
      "scoreCapability": 90,
      "scoreQuality": 89,
      "scoreReliability": 93,
      "scoreUsability": 87,
      "scoreIntegration": 93,
      "scoreTransparency": 91,
      "scoreGovernance": 94,
      "scoreCost": 88,
      "scoreFreshness": 92,
      "totalScore": 90.8,
      "rankCategory": 4,
      "testSummary": "Streamlined trace logging audited across PyTorch and Hugging Face pipelines.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/wandb-weave.pdf",
      "lastAssessed": "2026-10-08T16:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-promptfoo",
      "slug": "promptfoo-cli",
      "name": "Promptfoo Security & Eval",
      "categorySlug": "ai-infrastructure",
      "vendor": "Promptfoo OSS",
      "releaseLabel": "v0.98",
      "evidenceStatus": "tested",
      "scoreCapability": 92,
      "scoreQuality": 91,
      "scoreReliability": 91,
      "scoreUsability": 85,
      "scoreIntegration": 92,
      "scoreTransparency": 98,
      "scoreGovernance": 96,
      "scoreCost": 97,
      "scoreFreshness": 96,
      "totalScore": 92.7,
      "rankCategory": 5,
      "testSummary": "Red teaming probe execution verified: caught 94% prompt injection test vectors.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/promptfoo-redteam.log",
      "lastAssessed": "2026-10-10T09:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-helicone",
      "slug": "helicone-gateway",
      "name": "Helicone LLM Gateway & Cache",
      "categorySlug": "ai-infrastructure",
      "vendor": "Helicone",
      "releaseLabel": "v2.8",
      "evidenceStatus": "tested",
      "scoreCapability": 89,
      "scoreQuality": 88,
      "scoreReliability": 95,
      "scoreUsability": 91,
      "scoreIntegration": 96,
      "scoreTransparency": 92,
      "scoreGovernance": 91,
      "scoreCost": 93,
      "scoreFreshness": 94,
      "totalScore": 91.7,
      "rankCategory": 6,
      "testSummary": "Tested proxy latency overhead (<12ms) and semantic cache hit rate (42.5%).",
      "evidenceUrl": "https://aki1k.com/evidence/tests/helicone-proxy.log",
      "lastAssessed": "2026-10-07T14:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-perplexity",
      "slug": "perplexity-enterprise",
      "name": "Perplexity Enterprise Pro",
      "categorySlug": "ai-research",
      "vendor": "Perplexity AI",
      "releaseLabel": "v2026-Q3",
      "evidenceStatus": "verified",
      "scoreCapability": 90,
      "scoreQuality": 92,
      "scoreReliability": 89,
      "scoreUsability": 94,
      "scoreIntegration": 84,
      "scoreTransparency": 86,
      "scoreGovernance": 88,
      "scoreCost": 85,
      "scoreFreshness": 94,
      "totalScore": 89.8,
      "rankCategory": 1,
      "testSummary": "Audited citation accuracy: 96.1% primary-source linkage across 200 technical queries.",
      "evidenceUrl": "https://aki1k.com/evidence/research/perplexity-audit.log",
      "lastAssessed": "2026-10-11T10:30:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-elicit",
      "slug": "elicit-research",
      "name": "Elicit Research Assistant",
      "categorySlug": "ai-research",
      "vendor": "Elicit Inc",
      "releaseLabel": "v2026.2",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 93,
      "scoreReliability": 88,
      "scoreUsability": 91,
      "scoreIntegration": 82,
      "scoreTransparency": 94,
      "scoreGovernance": 89,
      "scoreCost": 88,
      "scoreFreshness": 93,
      "totalScore": 90.1,
      "rankCategory": 2,
      "testSummary": "Tested systematic review extraction on 100 PubMed RCT papers: 94.8% fidelity.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/elicit-pubmed.json",
      "lastAssessed": "2026-10-10T11:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-consensus",
      "slug": "consensus-app",
      "name": "Consensus Evidence Search",
      "categorySlug": "ai-research",
      "vendor": "Consensus AI",
      "releaseLabel": "v2026.1",
      "evidenceStatus": "verified",
      "scoreCapability": 89,
      "scoreQuality": 92,
      "scoreReliability": 89,
      "scoreUsability": 92,
      "scoreIntegration": 83,
      "scoreTransparency": 92,
      "scoreGovernance": 88,
      "scoreCost": 89,
      "scoreFreshness": 92,
      "totalScore": 89.7,
      "rankCategory": 3,
      "testSummary": "Audited Consensus Meter correlation with peer-reviewed meta-analyses at r=0.91.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/consensus-audit.pdf",
      "lastAssessed": "2026-10-09T13:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-scite",
      "slug": "scite-ai",
      "name": "Scite.ai Citation Context",
      "categorySlug": "ai-research",
      "vendor": "Research Solutions",
      "releaseLabel": "v2026.3",
      "evidenceStatus": "verified",
      "scoreCapability": 88,
      "scoreQuality": 91,
      "scoreReliability": 90,
      "scoreUsability": 89,
      "scoreIntegration": 85,
      "scoreTransparency": 95,
      "scoreGovernance": 90,
      "scoreCost": 87,
      "scoreFreshness": 91,
      "totalScore": 89.6,
      "rankCategory": 4,
      "testSummary": "Smart Citations classification audited across 50,000 supporting/contrasting citations.",
      "evidenceUrl": "https://aki1k.com/evidence/reviews/scite-smart-citations.json",
      "lastAssessed": "2026-10-08T10:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-semanticscholar",
      "slug": "semantic-scholar",
      "name": "Semantic Scholar AI Reader",
      "categorySlug": "ai-research",
      "vendor": "Allen Institute for AI",
      "releaseLabel": "v2026-oss",
      "evidenceStatus": "tested",
      "scoreCapability": 87,
      "scoreQuality": 90,
      "scoreReliability": 88,
      "scoreUsability": 88,
      "scoreIntegration": 86,
      "scoreTransparency": 97,
      "scoreGovernance": 92,
      "scoreCost": 98,
      "scoreFreshness": 90,
      "totalScore": 89.9,
      "rankCategory": 5,
      "testSummary": "Tested TLDR paper summarization accuracy against author abstract baselines.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/semantic-scholar-tldr.log",
      "lastAssessed": "2026-10-07T16:00:00Z",
      "methodologyVersion": "v1.0"
    },
    {
      "id": "prod-storm",
      "slug": "stanford-storm",
      "name": "Stanford STORM Research Agent",
      "categorySlug": "ai-research",
      "vendor": "Stanford OVAL",
      "releaseLabel": "v1.2-oss",
      "evidenceStatus": "tested",
      "scoreCapability": 91,
      "scoreQuality": 90,
      "scoreReliability": 85,
      "scoreUsability": 80,
      "scoreIntegration": 84,
      "scoreTransparency": 96,
      "scoreGovernance": 89,
      "scoreCost": 96,
      "scoreFreshness": 93,
      "totalScore": 89.3,
      "rankCategory": 6,
      "testSummary": "Tested autonomous Wikipedia-style synthesis from internet sources: 88.4% grounded citations.",
      "evidenceUrl": "https://aki1k.com/evidence/tests/storm-stanford.log",
      "lastAssessed": "2026-10-06T14:00:00Z",
      "methodologyVersion": "v1.0"
    }
  ],
  "change_history": [
    {
      "id": "chg-001",
      "productSlug": "cursor-ide",
      "productName": "Cursor IDE Agent",
      "categorySlug": "ai-tools",
      "changeType": "capability_gain",
      "previousValue": "Agent mode single-file edit limit",
      "newValue": "Full workspace multi-file synthesis",
      "impactDelta": 4.5,
      "evidenceUrl": "https://aki1k.com/evidence/changes/cursor-v045.diff",
      "detectedAt": "2026-10-08T12:00:00Z",
      "verifiedBy": "AKI Lab Core"
    },
    {
      "id": "chg-002",
      "productSlug": "windsurf-ide",
      "productName": "Windsurf IDE",
      "categorySlug": "ai-tools",
      "changeType": "capability_gain",
      "previousValue": "Standard file completion",
      "newValue": "Cascade flow autonomous context buffer",
      "impactDelta": 3.8,
      "evidenceUrl": "https://aki1k.com/evidence/changes/windsurf-cascade.diff",
      "detectedAt": "2026-10-06T15:00:00Z",
      "verifiedBy": "AKI Tool Audit"
    },
    {
      "id": "chg-003",
      "productSlug": "v0-vercel",
      "productName": "v0 Generative UI",
      "categorySlug": "ai-tools",
      "changeType": "pricing_shift",
      "previousValue": "Credit pack paywall for Next.js exports",
      "newValue": "Unlimited standard project exports on team tier",
      "impactDelta": 5.2,
      "evidenceUrl": "https://aki1k.com/evidence/changes/v0-pricing-2026.html",
      "detectedAt": "2026-10-04T10:00:00Z",
      "verifiedBy": "AKI Price Monitor"
    },
    {
      "id": "chg-004",
      "productSlug": "claude-code",
      "productName": "Claude Code CLI Agent",
      "categorySlug": "ai-agents",
      "changeType": "pricing_shift",
      "previousValue": "$0.03/1k input flat tier",
      "newValue": "$0.003/1k input prompt caching tier",
      "impactDelta": 8.2,
      "evidenceUrl": "https://aki1k.com/evidence/changes/claude-cache-announcement.html",
      "detectedAt": "2026-10-05T09:00:00Z",
      "verifiedBy": "AKI Price Monitor"
    },
    {
      "id": "chg-005",
      "productSlug": "cognition-devin",
      "productName": "Devin Software Engineer",
      "categorySlug": "ai-agents",
      "changeType": "governance_update",
      "previousValue": "Internal cloud sandbox only",
      "newValue": "Enterprise VPC isolated deployment",
      "impactDelta": 6.5,
      "evidenceUrl": "https://aki1k.com/evidence/changes/devin-enterprise-vpc.pdf",
      "detectedAt": "2026-10-02T11:00:00Z",
      "verifiedBy": "AKI Security Audit"
    },
    {
      "id": "chg-006",
      "productSlug": "swe-agent",
      "productName": "SWE-Agent Autonomous Bench",
      "categorySlug": "ai-agents",
      "changeType": "capability_gain",
      "previousValue": "Bash execution timeout at 30s",
      "newValue": "Subprocess streaming timeout at 120s with crash recovery",
      "impactDelta": 4,
      "evidenceUrl": "https://aki1k.com/evidence/changes/swe-agent-pr412.patch",
      "detectedAt": "2026-09-28T16:00:00Z",
      "verifiedBy": "AKI Open Benchmark"
    },
    {
      "id": "chg-007",
      "productSlug": "n8n-ai-nodes",
      "productName": "n8n AI Advanced Orchestration",
      "categorySlug": "ai-automation",
      "changeType": "governance_update",
      "previousValue": "Community unverified credential storage",
      "newValue": "SOC2 Type II verified enterprise vault",
      "impactDelta": 6,
      "evidenceUrl": "https://aki1k.com/evidence/changes/n8n-soc2.pdf",
      "detectedAt": "2026-10-01T15:00:00Z",
      "verifiedBy": "AKI Security Audit"
    },
    {
      "id": "chg-008",
      "productSlug": "make-automation",
      "productName": "Make AI Workflow Hub",
      "categorySlug": "ai-automation",
      "changeType": "regression_detected",
      "previousValue": "Zero data loss under burst webhook traffic",
      "newValue": "2.1% dropped webhooks observed during regional spike",
      "impactDelta": -3.2,
      "evidenceUrl": "https://aki1k.com/evidence/changes/make-incident-sep26.log",
      "detectedAt": "2026-09-25T14:30:00Z",
      "verifiedBy": "AKI Reliability Lab"
    },
    {
      "id": "chg-009",
      "productSlug": "activepieces-oss",
      "productName": "Activepieces Open Automation",
      "categorySlug": "ai-automation",
      "changeType": "capability_gain",
      "previousValue": "Sequential node execution only",
      "newValue": "Parallel branch execution with asynchronous promise joins",
      "impactDelta": 4.8,
      "evidenceUrl": "https://aki1k.com/evidence/changes/activepieces-v032.diff",
      "detectedAt": "2026-09-20T10:00:00Z",
      "verifiedBy": "AKI Automation Audit"
    },
    {
      "id": "chg-010",
      "productSlug": "claude-3-5-sonnet",
      "productName": "Claude 3.5 Sonnet",
      "categorySlug": "ai-models",
      "changeType": "capability_gain",
      "previousValue": "June 2024 checkpoint baseline",
      "newValue": "October 2024 refreshed checkpoint with Computer Use API",
      "impactDelta": 5.5,
      "evidenceUrl": "https://aki1k.com/evidence/changes/sonnet-oct-release.pdf",
      "detectedAt": "2026-10-02T08:00:00Z",
      "verifiedBy": "AKI Model Observatory"
    },
    {
      "id": "chg-011",
      "productSlug": "deepseek-v3",
      "productName": "DeepSeek-V3 Reasoning Core",
      "categorySlug": "ai-models",
      "changeType": "pricing_shift",
      "previousValue": "$0.55/1M token API pricing",
      "newValue": "$0.14/1M input and $0.28/1M output tokens",
      "impactDelta": 9.1,
      "evidenceUrl": "https://aki1k.com/evidence/changes/deepseek-pricing-v3.html",
      "detectedAt": "2026-10-04T12:00:00Z",
      "verifiedBy": "AKI Price Monitor"
    },
    {
      "id": "chg-012",
      "productSlug": "gpt-4o",
      "productName": "GPT-4o Multimodal",
      "categorySlug": "ai-models",
      "changeType": "regression_detected",
      "previousValue": "Strict tool-calling JSON schema adherence",
      "newValue": "Edge-case schema violation observed in nested array objects",
      "impactDelta": -2.4,
      "evidenceUrl": "https://aki1k.com/evidence/changes/gpt4o-json-regression.log",
      "detectedAt": "2026-09-22T17:00:00Z",
      "verifiedBy": "AKI Evaluation Core"
    },
    {
      "id": "chg-013",
      "productSlug": "langfuse-observability",
      "productName": "Langfuse LLM Observability",
      "categorySlug": "ai-infrastructure",
      "changeType": "capability_gain",
      "previousValue": "Synchronous span batch flush",
      "newValue": "Zero-blocking asynchronous ClickHouse ingestion",
      "impactDelta": 4.2,
      "evidenceUrl": "https://aki1k.com/evidence/changes/langfuse-clickhouse.patch",
      "detectedAt": "2026-09-30T13:00:00Z",
      "verifiedBy": "AKI Infra Observatory"
    },
    {
      "id": "chg-014",
      "productSlug": "promptfoo-cli",
      "productName": "Promptfoo Security & Eval",
      "categorySlug": "ai-infrastructure",
      "changeType": "capability_gain",
      "previousValue": "Static test assertions",
      "newValue": "LLM-assisted adaptive red teaming evaluation",
      "impactDelta": 5,
      "evidenceUrl": "https://aki1k.com/evidence/changes/promptfoo-redteam.diff",
      "detectedAt": "2026-10-03T18:00:00Z",
      "verifiedBy": "AKI Security Audit"
    },
    {
      "id": "chg-015",
      "productSlug": "helicone-gateway",
      "productName": "Helicone LLM Gateway & Cache",
      "categorySlug": "ai-infrastructure",
      "changeType": "pricing_shift",
      "previousValue": "Per-seat enterprise plan requirement",
      "newValue": "Free open-source self-hosting license ratified",
      "impactDelta": 6.8,
      "evidenceUrl": "https://aki1k.com/evidence/changes/helicone-oss-license.md",
      "detectedAt": "2026-09-24T11:00:00Z",
      "verifiedBy": "AKI Legal Audit"
    },
    {
      "id": "chg-016",
      "productSlug": "perplexity-enterprise",
      "productName": "Perplexity Enterprise Pro",
      "categorySlug": "ai-research",
      "changeType": "governance_update",
      "previousValue": "Standard web search logging",
      "newValue": "Zero data retention enterprise agreement guarantee",
      "impactDelta": 5.5,
      "evidenceUrl": "https://aki1k.com/evidence/changes/perplexity-enterprise-terms.pdf",
      "detectedAt": "2026-10-07T09:00:00Z",
      "verifiedBy": "AKI Security Audit"
    },
    {
      "id": "chg-017",
      "productSlug": "elicit-research",
      "productName": "Elicit Research Assistant",
      "categorySlug": "ai-research",
      "changeType": "capability_gain",
      "previousValue": "20-paper batch extraction limit",
      "newValue": "1,000-paper asynchronous systematic synthesis",
      "impactDelta": 6.2,
      "evidenceUrl": "https://aki1k.com/evidence/changes/elicit-batch1k.html",
      "detectedAt": "2026-09-29T15:00:00Z",
      "verifiedBy": "AKI Research Lab"
    },
    {
      "id": "chg-018",
      "productSlug": "consensus-app",
      "productName": "Consensus Evidence Search",
      "categorySlug": "ai-research",
      "changeType": "capability_gain",
      "previousValue": "Keyword search relevance",
      "newValue": "Academic semantic graph vector search with DOI verification",
      "impactDelta": 4.1,
      "evidenceUrl": "https://aki1k.com/evidence/changes/consensus-doi.diff",
      "detectedAt": "2026-09-26T12:00:00Z",
      "verifiedBy": "AKI Research Lab"
    }
  ]
}