🚀 Riksdagsmonitor — Future Flowchart Architecture

🔄 Process & Data-Flow Evolution: From Agentic Static Newsroom to AWS-Serverless Intelligence Platform
🎯 Build-Time OSINT · Party Landscape Analytics · Bedrock RAG · Predictive Democracy · Real-Time Fusion

📋 Document Owner: CEO | 📄 Version: 3.0 | 📅 Last Updated: 2026-05-31 (UTC)
🔄 Review Cycle: Quarterly | ⏰ Next Review: 2026-08-31
🏢 Owner: Hack23 AB (Org.nr 559534-7807) | 🏷️ Classification: Public

📚 Architecture Documentation Map

🎯 Executive Summary

This document maps the process and data-flow evolution of Riksdagsmonitor across three horizons (2026–2037). It is the forward-looking sibling of FLOWCHART.md: every future flow is built on top of the production v1.x baseline documented there — the autonomous agentic newsroom, the multi-provider data fetch/persist layer, the analysis-gate quality wall, and the multi-region static deploy.

The roadmap deliberately separates the static-deepening near term from the serverless long term, so investment stays grounded:

Strategic process vision:

• 🟢 Agentic newsroom hardening — keep zero-human-editor pipeline; tighten the analysis gate and provenance (2026)
• 🔵 Static OSINT precompute — move network/temporal/anomaly analytics to build time; render party-focused landscape dashboards & INTOP scorecards as static artifacts (2026–2027)
• 🟠 Bedrock runtime flows — multi-modal generation, real-time fact-checking and Knowledge-Bases RAG over the 109K+ document corpus (2028+)
• 🟠 Predictive pipelines — Step-Functions election & vote forecasting with calibrated uncertainty (2028+)
• 🟠 Streaming fusion — Kinesis + EventBridge real-time parliamentary monitoring and multi-source data fusion (2029+)
• 🟠 API economy & personalization — API Gateway public political-intelligence API + Cognito personalized feeds (2029+)
• 🛡️ Guardrails throughout — public data only, GDPR Art. 9(2)(e)/(g), party neutrality, no surveillance/psyops, human-on-the-loop governance per AI Policy

⚠️ Targets, not achievements. All metrics below the v1.x baseline are roadmap targets, explicitly labelled, never presented as already delivered.

📋 Table of Contents

1. Three-Horizon Process Evolution
2. v1.x Baseline Flows (Build the Future On This)
3. v2.0 Static-Deepening Flows (2026–2027)
4. v3.0+ AWS-Serverless Runtime Flows (2028–2037)
5. ISMS & AI-Policy Compliance Flows
6. AI Model Lifecycle & Evolution Flows
7. Performance & Cost-Optimization Flows
8. IMF Economic-Dataflow Evolution
9. Document Control
10. Related Documentation

1. 🧭 Three-Horizon Process Evolution

The platform's process architecture migrates along one axis — where computation happens — while the democratic-ethics guardrails stay constant. v1.x and v2.0 compute everything at build time and ship immutable static artifacts; v3.0+ moves compute to managed serverless runtime without ever introducing a server to patch.

flowchart LR
    subgraph H1["🟢 v1.x Baseline — Static Agentic Newsroom (2026)"]
        direction TB
        B1[gh-aw agentic workflows] --> B2[Analysis artifacts + gate]
        B2 --> B3[Article render + 14-lang translate]
        B3 --> B4[Multi-region static deploy]
    end

    subgraph H2["🔵 v2.0 — Static Deepening (2026–2027)"]
        direction TB
        V1[Build-time OSINT precompute<br/>network · temporal · anomaly] --> V2[Party-landscape dashboards]
        V2 --> V3[Source-grading + INTOP scorecards]
        V3 --> V4[Static HTML artifacts<br/>no runtime backend]
    end

    subgraph H3["🟠 v3.0+ — AWS Serverless (2028–2037)"]
        direction TB
        S1[API Gateway + Cognito] --> S2[Lambda + Step Functions]
        S2 --> S3[Bedrock + Knowledge Bases RAG]
        S3 --> S4[Kinesis streaming + EventBridge fusion]
    end

    H1 --> H2 --> H3

    style B1 fill:#4caf50,color:#000000
    style B4 fill:#4caf50,color:#000000
    style V1 fill:#2196f3,color:#ffffff
    style V4 fill:#2196f3,color:#ffffff
    style S1 fill:#ff9800,color:#000000
    style S4 fill:#ff9800,color:#000000
Copy

Horizon decision gates — promotion to the next horizon is conditional, not calendar-driven:

flowchart TD
    G0[v1.x in production] --> G1{Static precompute<br/>covers party + OSINT needs<br/>at acceptable build time?}
    G1 -->|Yes — stay static| G2[v2.0: deepen static OSINT]
    G1 -->|Build time > budget OR<br/>real-time demand proven| G3{Serverless ROI &<br/>governance ready?}
    G2 --> G3
    G3 -->|No| G2
    G3 -->|Yes — DPIA + cost model approved| G4[v3.0+: AWS serverless runtime]

    style G0 fill:#4caf50,color:#000000
    style G2 fill:#2196f3,color:#ffffff
    style G4 fill:#ff9800,color:#000000
    style G3 fill:#9c27b0,color:#ffffff
Copy

2. 🟢 v1.x Baseline Flows (Build the Future On This)

These flows are in production today and refreshed to match FLOWCHART.md. Every later horizon extends — never replaces — them.

2.1 Agentic Newsroom Pipeline (gh-aw → artifacts → article → deploy)

Objective: zero-human-editor publication of intelligence articles in 14 languages, gated by the analysis gate.

flowchart TD
    CRON[⏰ Scheduled gh-aw workflow<br/>14 agentic workflows] --> MCPI[Initialize MCP clients]
    MCPI --> FETCH[Multi-provider data fetch]

    FETCH --> RR[riksdag-regering MCP<br/>32+ tools]
    FETCH --> SCB[scb MCP — PxWeb v2]
    FETCH --> WB[world-bank MCP — WGI / non-econ]
    FETCH --> IMF[imf TS client via bash<br/>WEO / FM / IFS — primary economic]
    FETCH --> CIA[CIA platform JSON/CSV export]

    RR --> AGG[Context aggregation + dedup + relevance]
    SCB --> AGG
    WB --> AGG
    IMF --> AGG
    CIA --> AGG

    AGG --> ARTI[Produce analysis artifacts<br/>9 core / 14 Tier-C in analysis/daily/DATE/]
    ARTI --> GATE{Analysis Gate<br/>single blocking wall}
    GATE -->|Fail| BLOCK[Block + open issue<br/>no publication]
    GATE -->|Pass| GEN[Claude Opus 4.8<br/>article composition]
    GEN --> QC{Quality + evidence<br/>citations valid?}
    QC -->|No| RETRY[Refine prompt] --> GEN
    QC -->|Yes| RENDER[Template HTML render]
    RENDER --> TRANS[14-language translation<br/>Sonnet 4.6]
    TRANS --> SEO[Schema.org + hreflang + OG tags]
    SEO --> COMMIT[Git commit + signed provenance]
    COMMIT --> CI[CI/CD: HTMLHint · ESLint · Vitest · Cypress]
    CI --> DEPLOY[Multi-region static deploy]
    DEPLOY --> LIVE[📰 Live in 14 languages]

    style CRON fill:#2196f3,color:#ffffff
    style GATE fill:#ff006e,color:#ffffff
    style BLOCK fill:#f44336,color:#ffffff
    style GEN fill:#9c27b0,color:#ffffff
    style LIVE fill:#4caf50,color:#000000
Copy

2.2 Build-Time Political-Intelligence & Analysis-Gate Flow

Objective: horizon-stratified intelligence products precompiled before each static build.

flowchart TD
    T[generate-political-intelligence.ts] --> CAT[catalog.ts<br/>article-types.json registry]
    CAT --> STR[daily-streams.ts<br/>active streams per horizon]
    STR --> HZ{Horizon stratification}
    HZ --> H72[T+72h]
    HZ --> H7[T+7d]
    HZ --> H30[T+30d]
    HZ --> H90[T+90d]
    HZ --> H365[T+365d]
    HZ --> H1460[T+1460d election cycle]

    H72 --> GATE{Analysis Gate<br/>9 checks · artifact existence,<br/>evidence, Pass-2, cross-refs}
    H7 --> GATE
    H30 --> GATE
    H90 --> GATE
    H365 --> GATE
    H1460 --> GATE

    GATE -->|Fail| STOP[❌ Block pipeline]
    GATE -->|Pass| I18N[i18n/ — 14-language translation]
    I18N --> RND[render/ — template HTML]
    RND --> OUT[political-intelligence*.html<br/>14 variants]

    style T fill:#2196f3,color:#ffffff
    style GATE fill:#ff006e,color:#ffffff
    style STOP fill:#f44336,color:#ffffff
    style OUT fill:#4caf50,color:#000000
Copy

2.3 Data Fetch & Persist Flow (multi-provider, provenance-tagged)

flowchart LR
    TRIG[download-parliamentary-data.ts<br/>+ fetch-voting-records.ts<br/>+ fetch-calendar.ts] --> SRC{Source}
    SRC -->|Propositions / motions / betänkanden| RIKS[(data.riksdagen.se)]
    SRC -->|Government docs| G0V[(Regeringen via g0v.se)]
    SRC -->|Statistics| SCB[(SCB PxWeb v2)]
    SRC -->|Economic primary| IMF[(IMF SDMX 3.0 + Datamapper)]
    SRC -->|Governance / environment| WB[(World Bank)]
    SRC -->|Aggregated political engine| CIA[(CIA platform export)]

    RIKS --> VAL[Validate + transform]
    G0V --> VAL
    SCB --> VAL
    IMF --> VAL
    WB --> VAL
    CIA --> VAL
    VAL --> PROV[Attach provenance + freshness + SHA-256]
    PROV --> PERSIST[(data/ + analysis/data/ JSON + meta)]
    PERSIST --> READY[Ready for analysis + dashboard build]

    style TRIG fill:#2196f3,color:#ffffff
    style IMF fill:#0a4f8f,color:#ffffff
    style PERSIST fill:#ff9800,color:#000000
    style READY fill:#4caf50,color:#000000
Copy

2.4 Dashboard Build Flow (lazy-loaded TypeScript modules)

flowchart TD
    DATA[(Persisted JSON: 349 MPs · 2,494 historical ·<br/>3.5M+ votes · 109K+ docs)] --> TS[TypeScript dashboard modules<br/>Chart.js / D3.js]
    TS --> BUILD[Vite build — code-split per dashboard]
    BUILD --> LAZY[Lazy-loaded bundles<br/>~11 functional dashboards]
    LAZY --> A11Y[WCAG 2.1 AA pass + cyberpunk theme]
    A11Y --> STATIC[Static HTML + hashed assets]
    STATIC --> DEPLOY[Deploy step]

    style DATA fill:#ff9800,color:#000000
    style STATIC fill:#4caf50,color:#000000
    style A11Y fill:#9c27b0,color:#ffffff
Copy

2.5 Multi-Region Static Deploy & Disaster-Recovery Flow

flowchart TD
    MERGE[Merge to main] --> CIGATE[CI security gate<br/>CodeQL · Dependabot · SLSA · Sigstore]
    CIGATE --> OK{All gates green?}
    OK -->|No| HALT[Halt + alert]
    OK -->|Yes| S3P[(S3 primary — us-east-1)]
    S3P --> REPL[(S3 replica — eu-west-1)]
    S3P --> CF[CloudFront invalidation]
    REPL --> CF
    CF --> PROD[🌐 riksdagsmonitor.com live]
    PROD --> HEALTH{Origin healthy?}
    HEALTH -->|Degraded| DR[Fail over to GitHub Pages DR]
    HEALTH -->|Healthy| DONE[Serving from edge]
    DR --> DONE

    style MERGE fill:#2196f3,color:#ffffff
    style HALT fill:#f44336,color:#ffffff
    style PROD fill:#4caf50,color:#000000
    style DR fill:#ff9800,color:#000000
Copy

3. 🔵 v2.0 Static-Deepening Flows (2026–2027)

Strategic choice: stay static. v2.0 introduces no runtime backend. Every new capability is precomputed at build time and rendered into immutable HTML/CSS. The wins are deeper party analytics and higher-grade OSINT, not new infrastructure. AI generations of this window (Opus 4.6–4.9 → 5.x) primarily sharpen the agentic build pipeline.

3.1 Party-Focused Political-Landscape Dashboard Build

Objective: turn the persisted corpus into party-centric landscape dashboards — cohesion, coalition dynamics, bloc alignment, party-vs-party comparison, agenda tracking — all computed offline.

flowchart TD
    CORPUS[(Persisted corpus: votes · speeches · motions · members)] --> COH[Compute party-cohesion index<br/>Rice index per voting block]
    CORPUS --> BLOC[Compute bloc alignment<br/>government vs opposition matrices]
    CORPUS --> COAL[Coalition-dynamics model<br/>agreement / defection rates]
    CORPUS --> AGENDA[Agenda tracking<br/>topic share per party over time]

    COH --> AGG[Aggregate party-landscape dataset]
    BLOC --> AGG
    COAL --> AGG
    AGENDA --> AGG
    AGG --> NEU{Neutrality lint<br/>equal coverage per party?}
    NEU -->|Imbalanced| FIXN[Rebalance + flag] --> AGG
    NEU -->|Balanced| RENDER[Render party-landscape dashboards<br/>Chart.js / D3.js static]
    RENDER --> I18N[14-language labels]
    I18N --> OUT[Static party-landscape HTML]

    style CORPUS fill:#ff9800,color:#000000
    style NEU fill:#ff006e,color:#ffffff
    style OUT fill:#4caf50,color:#000000
    style AGG fill:#9c27b0,color:#ffffff
Copy

3.2 Build-Time OSINT Computation Pipeline (network · temporal · anomaly)

Objective: move structured tradecraft (network analysis, temporal/seasonal patterns, anomaly detection) from ad-hoc analysis into a repeatable build-time pipeline whose outputs are graded and rendered statically.

flowchart TD
    IN[(Corpus + provenance)] --> NET[Network analysis<br/>co-sponsorship + co-voting graph<br/>centrality · clustering · bridging]
    IN --> TEMP[Temporal analysis<br/>seasonal cadence · trend breaks]
    IN --> ANOM[Anomaly detection<br/>voting defection · attendance outliers]

    NET --> SCORE[Apply OSINT source-grading<br/>A–F reliability × 1–6 credibility]
    TEMP --> SCORE
    ANOM --> SCORE
    SCORE --> ACH{ACH check<br/>≥3 competing hypotheses<br/>per anomaly}
    ACH -->|Unsupported| DROP[Drop / downgrade claim]
    ACH -->|Supported with dok_id evidence| PROD[OSINT product dataset]
    PROD --> VIZ[Render network/temporal/anomaly dashboards]
    VIZ --> OUTO[Static OSINT HTML artifacts]

    style IN fill:#ff9800,color:#000000
    style SCORE fill:#9c27b0,color:#ffffff
    style ACH fill:#ff006e,color:#ffffff
    style DROP fill:#f44336,color:#ffffff
    style OUTO fill:#4caf50,color:#000000
Copy

3.3 Source-Grading & INTOP Scorecard Flow

Objective: produce evidence-graded INTOP (intelligence-operations) scorecards where every metric ties to a dok_id, vote count, or named actor.

flowchart LR
    CLAIM[Candidate claim / metric] --> EV{Evidence present?<br/>dok_id · vote count · named actor · primary URL}
    EV -->|No| REJ[Reject — generic statement]
    EV -->|Yes| GRADE[Assign source grade<br/>reliability + credibility]
    GRADE --> UNC[Attach uncertainty band]
    UNC --> CARD[Compose INTOP scorecard cell]
    CARD --> AUDIT[Append to data-source audit trail]
    AUDIT --> STATIC[Render static scorecard]

    style CLAIM fill:#2196f3,color:#ffffff
    style REJ fill:#f44336,color:#ffffff
    style GRADE fill:#9c27b0,color:#ffffff
    style STATIC fill:#4caf50,color:#000000
Copy

3.4 v2.0 Build Sequence (end-to-end, still no runtime backend)

sequenceDiagram
    participant W as gh-aw workflow
    participant D as Data fetch layer
    participant O as OSINT precompute
    participant P as Party-landscape compute
    participant G as Analysis gate
    participant R as Renderer (14 lang)
    participant S as Static deploy

    W->>D: Fetch + persist (provenance-tagged)
    D-->>W: Versioned JSON corpus
    W->>O: Run network/temporal/anomaly + source-grading
    W->>P: Compute cohesion/coalition/bloc/agenda
    O-->>G: OSINT products (graded)
    P-->>G: Party-landscape datasets (neutrality-checked)
    G->>G: 9-check validation + Pass-2 evidence
    alt Gate fails
        G-->>W: Block + open issue
    else Gate passes
        G->>R: Approved datasets
        R->>S: Static HTML/CSS artifacts
        S-->>W: Deployed to CloudFront + S3 (+ GH Pages DR)
    end
Copy

v2.0 outcome: richer intelligence, identical hosting model. Build time grows; attack surface does not. When build time exceeds budget or real-time demand is proven, the horizon-gate (§1) authorizes the v3.0 serverless move.

4. 🟠 v3.0+ AWS-Serverless Runtime Flows (2028–2037)

Strategic choice: all-in managed serverless. No Kubernetes, no containers, no servers to patch — only Lambda, Step Functions, EventBridge, Kinesis, Bedrock, API Gateway, Cognito, DynamoDB, Aurora Serverless v2, Neptune Serverless, OpenSearch Serverless, and Timestream. AWS Well-Architected aligned, multi-region resilient. AI generations Opus 6.x→AGI unlock conversational, predictive, and self-improving runtime flows.

4.1 Public Political-Intelligence API Request Flow (API Gateway · Lambda · Cognito)

flowchart TD
    USER[API consumer / citizen app] --> APIGW[Amazon API Gateway]
    APIGW --> AUTH{Cognito authorizer}
    AUTH -->|Anonymous public tier| RL1[Rate-limit: public quota]
    AUTH -->|Authenticated| RL2[Rate-limit: tiered quota]
    RL1 --> LMB[Lambda resolver]
    RL2 --> LMB
    LMB --> CACHE{Edge / DynamoDB cache hit?}
    CACHE -->|Yes| RESP[Return cached response]
    CACHE -->|No| DATA[Query Aurora SLv2 / Neptune / OpenSearch]
    DATA --> SHAPE[Shape + redact to public-data policy]
    SHAPE --> AUDIT[Emit access log → audit trail]
    AUDIT --> RESP
    RESP --> USER

    style USER fill:#2196f3,color:#ffffff
    style AUTH fill:#9c27b0,color:#ffffff
    style SHAPE fill:#ff006e,color:#ffffff
    style RESP fill:#4caf50,color:#000000
Copy

4.2 Bedrock Multi-Modal Generation + Real-Time Fact-Checking

flowchart TD
    EVT[EventBridge: parliamentary event] --> SF[Step Functions orchestration]
    SF --> EXTRACT[Lambda: extract structured data]
    EXTRACT --> BR[Amazon Bedrock<br/>multi-modal content generation]
    BR --> DRAFT[Draft article + media]
    DRAFT --> FC[Real-time fact-check Lambda]
    FC --> KB[(Bedrock Knowledge Bases<br/>RAG over corpus)]
    KB --> VERDICT{Claims grounded in<br/>primary sources?}
    VERDICT -->|Unsupported| CORRECT[Bedrock self-correction loop] --> FC
    VERDICT -->|Grounded| GUARD{Bedrock Guardrails<br/>neutrality + GDPR Art.9}
    GUARD -->|Violation| HOLD[Hold for human-on-the-loop review]
    GUARD -->|Pass| PUB[Publish + translate 14+ langs]
    PUB --> APIDX[Index into OpenSearch + Knowledge Base]

    style EVT fill:#2196f3,color:#ffffff
    style BR fill:#9c27b0,color:#ffffff
    style VERDICT fill:#ff006e,color:#ffffff
    style HOLD fill:#ff9800,color:#000000
    style PUB fill:#4caf50,color:#000000
Copy

4.3 Bedrock Knowledge-Bases RAG Semantic-Search Flow

sequenceDiagram
    participant U as User / agent
    participant GW as API Gateway
    participant L as Lambda (query)
    participant EMB as Bedrock embeddings
    participant OS as OpenSearch Serverless (vector)
    participant KB as Bedrock Knowledge Base
    participant FM as Bedrock foundation model

    U->>GW: Natural-language political query
    GW->>L: Authorized request (Cognito)
    L->>EMB: Embed query
    EMB-->>L: Query vector
    L->>OS: k-NN retrieve top passages (109K+ docs)
    OS-->>L: Ranked passages + dok_id citations
    L->>KB: Assemble grounded context
    KB->>FM: Generate answer with citations
    FM-->>L: Answer + source spans + uncertainty
    L-->>GW: Cited, neutral response
    GW-->>U: Answer (every claim → primary source)
Copy

4.4 Predictive Election & Vote-Forecasting Pipeline (Step Functions)

flowchart TD
    TRIG[Scheduled / event trigger] --> ING[Ingest features<br/>votes · polls · IMF macro · agenda]
    ING --> FE[Feature engineering Lambda]
    FE --> MODEL[Bedrock + custom models<br/>election & vote forecasting]
    MODEL --> CAL[Calibrate uncertainty<br/>scenario tree per horizon]
    CAL --> WEP{Word-of-estimative-probability<br/>ceiling per horizon band}
    WEP -->|Exceeds ceiling for horizon| DOWN[Downgrade confidence]
    WEP -->|Within band| SCEN[Emit 4-scenario + wildcard set]
    DOWN --> SCEN
    SCEN --> PIR[Roll forward PIRs to longer horizons]
    PIR --> STORE[(Timestream + DynamoDB)]
    STORE --> PUBF[Publish forecast products + API]

    style TRIG fill:#2196f3,color:#ffffff
    style MODEL fill:#9c27b0,color:#ffffff
    style WEP fill:#ff006e,color:#ffffff
    style PUBF fill:#4caf50,color:#000000
Copy

4.5 Real-Time Streaming & Multi-Source Data Fusion

flowchart LR
    SRC1[Riksdagen live feed] --> KIN[Amazon Kinesis streams]
    SRC2[Regeringen updates] --> KIN
    SRC3[SCB / IMF releases] --> KIN
    SRC4[CIA platform deltas] --> KIN
    KIN --> KDA[Kinesis Data Analytics<br/>windowed enrichment]
    KDA --> FUSE[Fusion Lambda<br/>entity resolution + dedup]
    FUSE --> EB[EventBridge bus]
    EB --> R1[Trigger fact-check]
    EB --> R2[Update forecasts]
    EB --> R3[Refresh dashboards / API cache]
    R1 --> STORE[(Aurora SLv2 · Neptune · OpenSearch · Timestream)]
    R2 --> STORE
    R3 --> STORE

    style KIN fill:#ff9800,color:#000000
    style FUSE fill:#9c27b0,color:#ffffff
    style EB fill:#2196f3,color:#ffffff
    style STORE fill:#4caf50,color:#000000
Copy

4.6 Personalized Feed Flow (Cognito identity, privacy-preserving)

flowchart TD
    LOGIN[User sign-in via Cognito] --> CONSENT{Explicit consent<br/>GDPR Art. 9}
    CONSENT -->|Declined| ANON[Serve neutral default feed]
    CONSENT -->|Granted| PREF[Store minimal preferences<br/>data minimisation]
    PREF --> RECO[Bedrock recommendation<br/>topic / committee / region]
    RECO --> NEU{Neutrality filter<br/>no partisan amplification}
    NEU -->|Skew detected| REBAL[Rebalance feed] --> RECO
    NEU -->|Balanced| FEED[Personalized intelligence feed]
    ANON --> FEED
    FEED --> RIGHTS[Honour access / erasure rights]

    style LOGIN fill:#2196f3,color:#ffffff
    style CONSENT fill:#ff006e,color:#ffffff
    style NEU fill:#ff006e,color:#ffffff
    style FEED fill:#4caf50,color:#000000
Copy

4.7 Continuous Model Improvement + Federated Learning

flowchart TD
    SIG[Signals: corrections · gate fails · user feedback] --> EVAL[Bedrock model evaluation harness]
    EVAL --> FED{Federated / privacy-preserving<br/>training applicable?}
    FED -->|Yes| LOCAL[Local gradient aggregation<br/>differential privacy noise]
    FED -->|No| CENTRAL[Centralised eval set<br/>public data only]
    LOCAL --> CAND[Candidate model]
    CENTRAL --> CAND
    CAND --> SHADOW[Shadow run 7 days<br/>parallel to production]
    SHADOW --> QG{Quality gate pass?}
    QG -->|No| ROLLBACK[Retain current model]
    QG -->|Yes| ROLLOUT[Gradual rollout 5%→25%→100%]
    ROLLOUT --> DOC[Update model-card + docs]

    style SIG fill:#2196f3,color:#ffffff
    style FED fill:#ff006e,color:#ffffff
    style SHADOW fill:#9c27b0,color:#ffffff
    style ROLLOUT fill:#4caf50,color:#000000
    style ROLLBACK fill:#ff9800,color:#000000
Copy

4.8 Crowdsourced Fact-Checking Flow (human-in-the-loop)

flowchart TD
    PUBL[Published claim] --> SUBMIT[Contributor submits challenge<br/>via Cognito-auth portal]
    SUBMIT --> SPAM{Anti-abuse + reputation check}
    SPAM -->|Reject| DENY[Discard]
    SPAM -->|Accept| TRIAGE[Bedrock triage<br/>cluster + evidence-match]
    TRIAGE --> EVID{Primary-source evidence?}
    EVID -->|No| CLOSE[Close — insufficient evidence]
    EVID -->|Yes| REVIEW[Human-on-the-loop adjudication]
    REVIEW --> DECIDE{Upheld?}
    DECIDE -->|Yes| CORRECT[Issue correction + audit entry]
    DECIDE -->|No| EXPLAIN[Document rationale]
    CORRECT --> KBUPD[Update Knowledge Base + reputation]

    style PUBL fill:#2196f3,color:#ffffff
    style EVID fill:#ff006e,color:#ffffff
    style REVIEW fill:#9c27b0,color:#ffffff
    style CORRECT fill:#4caf50,color:#000000
    style DENY fill:#f44336,color:#ffffff
Copy

4A. 🛰️ Political-Intelligence Cycle Process Flows (OSINT/INTOP, to 2037)

These flows realize the Political-Intelligence Capability Catalog (C1–C32) as the intelligence cycle — Direction → Collection → Processing → Analysis → Production → Dissemination → Feedback. The current platform runs a strong single-source, build-time, document-centric version of this cycle; the flows below add the fusion, warning, continuous-calibration and counter-AI loops that are missing today. Every loop keeps a human-on-the-loop checkpoint, an evidence/provenance anchor, and a party-symmetry/neutrality gate before anything is published.

4A.1 Master intelligence-cycle loop

flowchart LR
    DIR[🎯 Direction<br/>PIRs and tripwires] --> COL[📡 Collection<br/>multi-source OSINT]
    COL --> PROC[⚙️ Processing<br/>entity-resolve plus provenance]
    PROC --> FUSE[🔗 Fusion<br/>multi-INT graph]
    FUSE --> ANAL[🧠 Analysis<br/>SAT plus forecasting]
    ANAL --> PROD[📝 Production<br/>estimative products]
    PROD --> NEUT{Neutrality plus<br/>evidence gate}
    NEUT -->|Fail| ANAL
    NEUT -->|Pass| DISS[📤 Dissemination<br/>COP plus briefings]
    DISS --> FB[🔄 Feedback<br/>calibration plus retasking]
    FB --> DIR

    style DIR fill:#9c27b0,color:#ffffff
    style FUSE fill:#2196f3,color:#ffffff
    style NEUT fill:#ff006e,color:#ffffff
    style DISS fill:#4caf50,color:#000000
    style FB fill:#ff9800,color:#000000
Copy

4A.2 Multi-INT fusion & entity-resolution flow (C1 / C6)

flowchart TD
    SRC[Public sources<br/>riksdag plus registers plus discourse] --> NORM[Normalize plus stamp provenance]
    NORM --> ER{Entity match?<br/>deterministic plus embedding}
    ER -->|Ambiguous| HOLD[Hold for human review]
    ER -->|Resolved| CANON[Canonical entity ID]
    CANON --> EDGE[Build fusion edge<br/>plus Admiralty grade]
    EDGE --> GRADE{Evidence-anchored?<br/>dok_id present}
    GRADE -->|No| DROP[Reject — no analytic record]
    GRADE -->|Yes| GRAPH[Multi-INT fusion graph]
    GRAPH --> SAL[Salience plus neutrality note]
    SAL --> COP[Common Operating Picture]

    style SRC fill:#2196f3,color:#ffffff
    style ER fill:#ff006e,color:#ffffff
    style GRADE fill:#ff006e,color:#ffffff
    style DROP fill:#f44336,color:#ffffff
    style COP fill:#4caf50,color:#000000
Copy

4A.3 Indications & Warning tripwire flow (C14)

flowchart TD
    STREAM[Continuous public-data stream] --> IND[Compute indicators<br/>cohesion, motions, calendar]
    IND --> TW{Threshold breached?}
    TW -->|No| STREAM
    TW -->|Yes| LEVEL[Set warning level<br/>watch / elevated / high]
    LEVEL --> WEP[Attach calibrated probability<br/>ICD-203 lexicon]
    WEP --> HUM{Human triage}
    HUM -->|False alarm| TUNE[Tune threshold plus log]
    HUM -->|Confirm| RETASK[Re-task fusion plus forecasting]
    RETASK --> BRIEF[Warning note to watchers]
    TUNE --> STREAM

    style STREAM fill:#2196f3,color:#ffffff
    style TW fill:#ff006e,color:#ffffff
    style HUM fill:#9c27b0,color:#ffffff
    style BRIEF fill:#4caf50,color:#000000
Copy

4A.4 Forecasting & continuous-calibration loop (C13 / C29)

flowchart TD
    PIR[Forecasting question / PIR] --> ASSUME[Key-assumptions check]
    ASSUME --> MODEL[Ensemble forecast<br/>stats plus LLM scenarios]
    MODEL --> WEPF[Express as WEP plus point p]
    WEPF --> PUBF[Publish forecast plus assumptions]
    PUBF --> WAIT[Await resolution]
    WAIT --> RES{Event resolved?}
    RES -->|No| WAIT
    RES -->|Yes| SCORE[Brier-score plus update ledger]
    SCORE --> CAL{Calibration drifting?}
    CAL -->|Yes| RETRAIN[Retrain plus re-weight ensemble]
    CAL -->|No| PIR
    RETRAIN --> PIR

    style PIR fill:#9c27b0,color:#ffffff
    style RES fill:#ff006e,color:#ffffff
    style SCORE fill:#2196f3,color:#ffffff
    style CAL fill:#ff006e,color:#ffffff
    style RETRAIN fill:#ff9800,color:#000000
Copy

4A.5 FIMI / influence early-warning flow (C20) — aggregate, non-accusatory

flowchart TD
    DISC[Aggregate public discourse] --> FRAME[Detect narrative frames<br/>DISARM TTPs]
    FRAME --> COORD{Coordination signal?<br/>network-level only}
    COORD -->|No| LOG[Log frame, no alert]
    COORD -->|Yes| ATTR[Attribution confidence<br/>low/med, WEP-qualified]
    ATTR --> ETH{Ethics gate<br/>no citizen profiling}
    ETH -->|Fail| HALT[Halt — privacy breach risk]
    ETH -->|Pass| ADV[Advisory signal to analysts]
    ADV --> HUM[Human-on-the-loop framing]
    HUM --> BRIEFF[Contextual briefing, not accusation]

    style DISC fill:#2196f3,color:#ffffff
    style COORD fill:#ff006e,color:#ffffff
    style ETH fill:#ff006e,color:#ffffff
    style HALT fill:#f44336,color:#ffffff
    style BRIEFF fill:#4caf50,color:#000000
Copy

4A.6 Counter-AI & analytic-integrity gate (C26–C32)

flowchart TD
    IN[AI-assisted analytic step] --> INJ{Prompt-injection /<br/>poisoning screen}
    INJ -->|Detected| QUAR[Quarantine input plus alert]
    INJ -->|Clean| PROV{Provenance plus C2PA valid?}
    PROV -->|No| REFUSE[Refuse to cite]
    PROV -->|Yes| SAT[SAT-assisted reasoning<br/>ACH plus devil's advocate]
    SAT --> SYM{Party-symmetry audit}
    SYM -->|Asymmetric| REWORK[Rebalance plus re-review]
    SYM -->|Symmetric| ICD{ICD-203 compliance?}
    ICD -->|No| REWORK
    ICD -->|Yes| SIGN[Human sign-off plus publish]

    style IN fill:#2196f3,color:#ffffff
    style INJ fill:#ff006e,color:#ffffff
    style PROV fill:#ff006e,color:#ffffff
    style SYM fill:#ff006e,color:#ffffff
    style REFUSE fill:#f44336,color:#ffffff
    style SIGN fill:#4caf50,color:#000000
Copy

5. 🛡️ ISMS & AI-Policy Compliance Flows

Compliance is a gate in every horizon, mapped to ISO 27001:2022, NIST CSF 2.0 and CIS Controls v8.1, under Hack23 ISMS-PUBLIC.

5.1 AI-Policy Compliance Workflow (all horizons)

flowchart TD
    START[AI-assisted process invoked] --> PUB{Public data only?}
    PUB -->|No| REJECT[Reject — policy violation]
    PUB -->|Yes| LAWFUL{"GDPR Art. 9 lawful basis<br/>9(2)(e) public / 9(2)(g) public interest"}
    LAWFUL -->|None| REJECT
    LAWFUL -->|Established| NEUTRAL{Party neutrality maintained?}
    NEUTRAL -->|Skewed| REBAL[Rebalance + document] --> NEUTRAL
    NEUTRAL -->|Balanced| EVID{Evidence standard met?<br/>dok_id / vote / named actor}
    EVID -->|No| BLOCKE[Block — generic claim]
    EVID -->|Yes| HITL{High-risk processing?}
    HITL -->|Yes| DPIA[Run DPIA + human-on-the-loop]
    HITL -->|No| LOG[Log decision + provenance]
    DPIA --> LOG
    LOG --> APPROVE[✅ Proceed]

    style START fill:#2196f3,color:#ffffff
    style REJECT fill:#f44336,color:#ffffff
    style BLOCKE fill:#f44336,color:#ffffff
    style DPIA fill:#9c27b0,color:#ffffff
    style APPROVE fill:#4caf50,color:#000000
Copy

5.2 Control Mapping (process → framework)

6. 🤖 AI Model Lifecycle & Evolution Flows

6.1 AI Model Lifecycle Management (continuous evaluation → rollout)

graph TD
    subgraph EVAL["Continuous Model Evaluation (~quarterly cadence)"]
        A[New model release<br/>Opus 4.8 → 5.x → 6.x …] --> B{Benchmark vs current}
        B -->|Superior| C[Shadow test — 7-day parallel run]
        B -->|Equal/Inferior| D[Document + keep current]
        C --> E{Quality gate pass?}
        E -->|Yes| F[Gradual rollout 5%→25%→100%]
        E -->|No| G[Rollback + retain current]
        F --> H[Full deploy + update model-card]
    end

    subgraph MAJOR["Annual Major Upgrade"]
        I[Major version<br/>Opus 5.0 / 6.0 / 7.0 …] --> J{Architecture compatible?}
        J -->|Yes| K[Enable new capabilities]
        J -->|No| L[Adapt platform + re-test]
        K --> M[Integration test — 14+ languages]
        L --> M
        M --> N[Deploy behind feature flags]
    end

    subgraph COMP["Competitor Evaluation (quarterly)"]
        O[Review OpenAI · Google · Meta · EU sovereign AI] --> P{Better per-task model?}
        P -->|Yes| Q[Multi-model via Amazon Bedrock]
        P -->|No| R[Keep provider strategy]
        Q --> S[A/B test per task type]
    end

    subgraph PARA["Paradigm-Shift Watch (2031–2037)"]
        U[Quantum / neuromorphic / AGI signal] --> V{Paradigm shift viable?}
        V -->|Yes| W[Autonomous mode + human oversight]
        V -->|Partial| X[Enhanced mode]
        V -->|No| Y[Continue annual upgrades]
    end

    H --> I
    N --> O
    S --> U

    style A fill:#00d9ff,color:#000000
    style I fill:#ff006e,color:#ffffff
    style O fill:#ffbe0b,color:#000000
    style U fill:#9c27b0,color:#ffffff
Copy

6.2 AI Model Evolution Timeline — DevSecOps Capability (verbatim)

Assumptions: major AI-model upgrades land roughly annually; competitors (OpenAI, Google, Meta, EU sovereign AI) are re-evaluated at each release; the architecture is built to absorb paradigm shifts (quantum AI, neuromorphic computing); all transitions are governed by the Hack23 AI Policy.

6.3 The Same AI Curve, Translated into Process Terms

Each model generation unlocks a concrete change in where and how the platform processes political intelligence:

flowchart LR
    Y26[2026 Opus 4.x<br/>🟢 Static agentic newsroom] --> Y27[2027 Opus 5.x<br/>🔵 Static OSINT precompute]
    Y27 --> Y28[2028 Opus 6.x<br/>🟣 Bedrock RAG + fact-check]
    Y28 --> Y29[2029 Opus 7.x<br/>🟠 Streaming fusion]
    Y29 --> Y30[2030 Opus 8.x<br/>🔴 Calibrated forecasting]
    Y30 --> Y33[2031–33 Pre-AGI<br/>⚪ Nordic/EU federation]
    Y33 --> Y37[2034–37 AGI<br/>⭐ Transformative flows]

    style Y26 fill:#4caf50,color:#000000
    style Y27 fill:#2196f3,color:#ffffff
    style Y28 fill:#9c27b0,color:#ffffff
    style Y29 fill:#ff9800,color:#000000
    style Y30 fill:#f44336,color:#ffffff
    style Y33 fill:#eceff1,color:#000000
    style Y37 fill:#ffd700,color:#000000
Copy

7. ⚡ Performance & Cost-Optimization Flows

All numbers are roadmap targets, not achieved metrics.

7.1 Performance Targets by Horizon

7.2 Cost-Optimization Decision Flow

flowchart TD
    REQ[New capability requested] --> STATIC{Can it be precomputed<br/>at build time?}
    STATIC -->|Yes| BUILD[Render static — near-zero runtime cost]
    STATIC -->|No| RUNTIME{Real-time demand proven?}
    RUNTIME -->|No| DEFER[Defer — keep static]
    RUNTIME -->|Yes| SERVERLESS[Serverless — pay-per-use Lambda]
    SERVERLESS --> CACHE[Cache aggressively<br/>edge + DynamoDB TTL]
    CACHE --> BATCH[Batch Bedrock calls off-peak]
    BATCH --> BUDGET{Within cost budget?}
    BUDGET -->|No| TUNE[Tune model tier / quotas] --> BUDGET
    BUDGET -->|Yes| SHIP[Ship + monitor cost telemetry]

    style REQ fill:#2196f3,color:#ffffff
    style BUILD fill:#4caf50,color:#000000
    style DEFER fill:#ff9800,color:#000000
    style SHIP fill:#4caf50,color:#000000
Copy

Optimization levers: static-first default · CloudFront cache (99% hit-rate target) · Bedrock model-tier selection per task · off-peak batch generation · Brotli/gzip compression · serverless scale-to-zero · vintage-aware data caching (avoid redundant IMF/SCB fetches).

8. 🌐 IMF Economic-Dataflow Evolution

Baseline: the already-implemented IMF pipeline is documented in FLOWCHART.md. The diagram below layers future gates (vintage-age UI badge, provider-mix telemetry, RAG-indexed economic provenance) onto today's pure-TypeScript client without changing the canonical IMF-first rule.

Authoritative hub: analysis/imf/README.md · analysis/imf/agentic-integration.md · analysis/imf/indicators-inventory.json · analysis/imf/data-dictionary.md · .github/aw/ECONOMIC_DATA_CONTRACT.md v2.1

flowchart LR
    classDef primary fill:#0a4f8f,color:#fff,stroke:#00d9ff,stroke-width:2px
    classDef secondary fill:#3a3a3a,color:#ddd,stroke:#888
    classDef gate fill:#ff006e,color:#fff,stroke:#fff
    classDef future fill:#9c27b0,color:#fff,stroke:#fff

    Start([news-* / forecast workflow]) --> Domain{Identify economic class}
    Domain -->|Macro · Fiscal · Monetary · External · Trade| IMF[(IMF SDMX 3.0 + Datamapper<br/>scripts/imf-client.ts)]:::primary
    Domain -->|Governance / Environment / Social residue| WB[(World Bank)]:::secondary
    Domain -->|Swedish monthly / regional| SCB[(SCB PxWeb v2)]:::secondary

    IMF --> Vintage{Vintage > 6 months?}:::gate
    Vintage -->|Yes| Annotate[Annotate stale + downgrade confidence]
    Vintage -->|No| Cache[Cache: vintage-tagged · SHA-256 pinned]
    Annotate --> Cache
    Cache --> Provenance[Emit economicProvenance<br/>provider · dataflow · indicator · vintage]
    WB --> Cache
    SCB --> Cache

    Provenance --> Compose[Article / forecast composition]
    Compose --> Lint{IMF-first lint}:::gate
    Lint -->|WB economic citation w/o IMF cross-ref| Reject([Block — open issue])
    Lint -->|Pass| Publish([Publish])

    Publish --> FutBadge[v2.0: vintage-age UI badge<br/>+ provider-mix telemetry]:::future
    Publish --> FutRAG[v3.0+: index economicProvenance<br/>into Bedrock Knowledge Base]:::future

    style Start fill:#2196f3,color:#ffffff
    style Publish fill:#4caf50,color:#000000
Copy

Provider decision matrix

Canonical rule (unchanged across horizons). Every economic claim cites an IMF dataflow first; World Bank citations are reserved for governance, environment and social residue; SCB is the Swedish-specific ground-truth layer. In v3.0+, this same provenance is indexed into the Bedrock Knowledge Base so RAG answers inherit the IMF-first discipline. See ECONOMIC_DATA_CONTRACT.md v2.1 for the banned-phrase list and vintage discipline (>6 mo → annotation).

📝 Document Control

Version History:

Review Schedule:

• Quarterly review aligned with FUTURE_ARCHITECTURE.md and FUTURE_SECURITY_ARCHITECTURE.md
• Updated as capabilities reach implementation milestones and horizon-gates are crossed

Classification: Public
Distribution: Unrestricted
Repository: https://github.com/Hack23/riksdagsmonitor
Path: /FUTURE_FLOWCHART.md
Format: Markdown with Mermaid diagrams
Next Review: 2026-08-31

📚 Related Documentation

Current State (baseline this doc builds on)

• FLOWCHART.md — current process flows & pipelines (direct baseline)
• ARCHITECTURE.md — current system architecture (C4 model)
• DATA_MODEL.md — current data structures
• STATEDIAGRAM.md — current state transitions
• WORKFLOWS.md — CI/CD automation
• SECURITY_ARCHITECTURE.md — current security controls

Future Vision (sibling FUTURE_* docs)

• FUTURE_ARCHITECTURE.md — architectural evolution roadmap
• FUTURE_DATA_MODEL.md — enhanced data architecture
• FUTURE_STATEDIAGRAM.md — advanced state management
• FUTURE_MINDMAP.md — future capability map
• FUTURE_SWOT.md — strategic outlook
• FUTURE_SECURITY_ARCHITECTURE.md — security evolution roadmap

ISMS & Compliance

• Hack23 ISMS-PUBLIC — Information Security Management System
• AI Policy — AI governance and ethics
• Secure Development Policy — SDLC security
• Classification Framework — data and system classification

CIA Platform & Standards

• CIA Platform — Citizen Intelligence Agency (data source)
• EU AI Act · GDPR · ISO 27001 · NIST AI RMF

🌐 Riksdagsmonitor — Building the Future of Democratic Transparency
Powered by AI, Grounded in Privacy, Committed to Democracy

Website · GitHub · CIA Platform · ISMS

📋 Document Control:
✅ Approved by: James Pether Sörling, CEO
📤 Distribution: Public
🏷️ Classification:
📅 Effective Date: 2026-05-31
⏰ Next Review: 2026-08-31
🎯 Framework Compliance:

🔗 Hack23 Ecosystem

🗳️ Empower citizens · 🔍 Strengthen democratic accountability · 🕵️ Illuminate the political process

_{© 2008–2026 Hack23 AB (Org.nr 559534-7807) · Maintainer: James Pether Sörling, CISSP CISM}