🚀 Riksdagsmonitor — Future Mindmap

🗺️ Capability-Expansion Conceptual Map for Democratic Intelligence Evolution
🎯 Static v1.x · Party-Focused v2.0 · AWS-Serverless & Bedrock v3.0+ · Nordic & EU Expansion

📋 Document Owner: CEO | 📄 Version: 3.1 | 📅 Last Updated: 2026-05-31 (UTC)
🔄 Review Cycle: Quarterly | ⏰ Next Review: 2026-08-31
🏢 Owner: Hack23 AB (Org.nr 559534-7807) | 🏷️ Classification: Public

---

🎯 Purpose

This document provides future-state capability-expansion mindmaps for Riksdagsmonitor, visualizing how the platform's conceptual surface grows from a static Swedish-Parliament intelligence website into a serverless, AI-native democratic-intelligence platform. It builds directly on the freshly-refreshed current-state Mindmap and structures conceptual growth across three horizons spanning the 10-year look-ahead 2026–2037:

1. v1.x baseline — the accurate current state we build upon: 14-language static HTML/CSS, autonomous AI newsroom, CIA-platform + open-data integration, structured tradecraft (ACH, SWOT, PESTLE, STRIDE).
2. v2.0 (2026–2027) static — go deeper without migrating: party-focused political-landscape dashboards, advanced OSINT/INTOP quality, richer multilingual intelligence products, accessibility and performance — all on the retained static architecture.
3. v3.0+ (2028–2037) AWS serverless — the strategic platform leap: Amazon Bedrock conversational intelligence, Bedrock Knowledge Bases / RAG, API Gateway public API economy, Cognito personalization, predictive analytics, semantic search & knowledge graph, Nordic + EU parliament expansion, and real-time streaming intelligence.

Every roadmap claim is tied to a concrete capability, a named AWS service, an MCP server, or a dok-level data source. Future targets are labelled targets, never claimed as already achieved. All horizons preserve the platform's democratic-ethics, GDPR Article 9, neutrality and public-data-only guardrails.

🔭 Companion forward-looking docs: Future Architecture · Future Data Model · Future Flowchart · Future State Diagram · Future SWOT · Future Security Architecture

📋 Executive Summary

Riksdagsmonitor's conceptual surface expands along three horizons that separate analytical depth from architectural substrate:

• 🟢 Horizon 1 (v1.x, shipped) — a 14-language static HTML/CSS intelligence site with an autonomous AI newsroom (14 gh-aw workflows, Claude Opus 4.8 + Sonnet 4.6), CIA-platform and open-data integration, and structured tradecraft (ACH, SWOT, PESTLE, STRIDE). This is the proven base, not a placeholder.
• 🔵 Horizon 2 (v2.0, 2026–2027) — keep static, go deeper. The architecture does not migrate. Investment concentrates on party-focused landscape dashboards (cohesion, coalition dynamics, bloc alignment, party-vs-party), advanced OSINT/INTOP quality (network/temporal/geospatial patterns, anomaly detection, source-graded evidence, INTOP scorecards), richer multilingual products, and WCAG 2.2 / performance. AI stays a build-time author.
• 🟣 Horizon 3 (v3.0+, 2028–2037) — all-in AWS serverless. AI becomes a run-time interlocutor: Amazon Bedrock conversational intelligence and Agents, Bedrock Knowledge Bases / RAG over the 109k-document corpus, API Gateway public API economy, Cognito personalization, predictive analytics/forecasting, semantic search with a Neptune knowledge graph, Nordic + EU parliament expansion, and Kinesis real-time streaming — with zero infrastructure to manage.

Four invariants survive every horizon: public-data-only, political neutrality, GDPR Article 9 lawful bases (9(2)(e), 9(2)(g)), and Hack23 ISMS alignment (ISO 27001 / NIST CSF 2.0 / CIS v8.1). The AI Model Evolution curve (2026–2037) is mapped not only to DevSecOps but to concrete intelligence unlocks at each generation. All future figures are targets, never claimed as achieved.

🆕 v3.1 — post-2.0+ closure pass. This revision was produced as a final review against SWOT.md to guarantee every weakness, opportunity and improvement area has an explicit home on the future mindmaps. It adds three sections: an Agentic AI & Autonomous Multi-Agent Operations theme that maps what future AI agents really make possible (closing W1/W5 and amplifying O7), a Sustainability, Growth & Audience Expansion theme (closing W2/W3 and opening O4/O5/O6), and a fully-traceable SWOT → Future Coverage Crosswalk that confirms all 27 SWOT items (S1–S10, W1–W5, O1–O7, T1–T5) are addressed across the three horizons.

📚 Architecture Documentation Map

Document	Focus	Description
🏛️ Architecture	🏗️ C4 Models	System context, containers, components
📊 Data Model	📊 Data	Entity relationships and data dictionary
🔄 Flowchart	🔄 Processes	Business and data flow diagrams
📈 State Diagram	📈 States	System state transitions and lifecycles
🧠 Mindmap	🧠 Concepts	System conceptual relationships
💼 SWOT	💼 Strategy	Strategic analysis and positioning
🛡️ Security Architecture	🔒 Security	Current security controls and design
🚀 Future Security	🔮 Security	Planned security improvements
🎯 Threat Model	🎯 Threats	STRIDE/MITRE ATT&CK analysis
🚀 Future Architecture	🔮 Evolution	Architectural evolution roadmap
📊 Future Data Model	🔮 Data	Enhanced data architecture plans
🔄 Future Flowchart	🔮 Processes	Improved process workflows
📈 Future State Diagram	🔮 States	Advanced state management
🧠 Future Mindmap	🔮 Concepts	Capability expansion plans (this doc)
💼 Future SWOT	🔮 Strategy	Future strategic opportunities

---

📑 Table of Contents

1. Three-Horizon Capability Evolution (master map)
2. Horizon 1 — v1.x Baseline Capabilities
3. Horizon 2 — v2.0 Static Deepening (2026–2027)
4. Horizon 3 — v3.0+ AWS Serverless (2028–2037)
5. Theme — Data & Sources
6. Theme — OSINT / INTOP Tradecraft
7. Theme — Party & Coalition Analytics
8. Theme — AI / Bedrock Capabilities
9. Theme — Platform / AWS Serverless
10. Theme — Multilingual & Accessibility
11. Theme — Governance / Ethics / GDPR
12. Theme — Ecosystem & Expansion
13. AI / LLM Capability-Evolution (2026–2037)
14. Political-Intelligence Capability Catalog (to 2037) — master OSINT/INTOP map
15. Agentic AI & Autonomous Multi-Agent Operations
16. Sustainability, Growth & Audience Expansion
17. SWOT → Future Coverage Crosswalk
18. Capability Matrix (horizon-aligned)
19. Stakeholder Value & Roadmap Risk
20. IMF Data-Sources Future Surface

---

1. 🗺️ Three-Horizon Capability Evolution (master map)

The master mindmap frames the whole roadmap as three concentric horizons. Horizon 1 is shipped (v1.x); Horizon 2 deepens the static platform; Horizon 3 is the strategic AWS-serverless leap. Read inner-to-outer as time and capability both expand.

mindmap
  root((🚀 Riksdagsmonitor<br/>Capability Horizons))
    🟢 Horizon 1 v1.x Baseline
      Static HTML CSS 14 languages
      Autonomous AI newsroom gh-aw
      CIA platform plus open data
      Structured tradecraft ACH SWOT PESTLE STRIDE
      CloudFront S3 multi-region plus Pages DR
    🔵 Horizon 2 v2.0 Static 2026 to 2027
      Party-focused landscape dashboards
      Advanced OSINT INTOP quality
      Richer multilingual intelligence products
      Accessibility WCAG 2.2 and performance
      Static architecture retained no migration
    🟣 Horizon 3 v3.0 plus Serverless 2028 to 2037
      Amazon Bedrock conversational intelligence
      Bedrock Knowledge Bases RAG over corpus
      API Gateway public API economy
      Cognito identity and personalization
      Predictive analytics and forecasting
      Semantic search and knowledge graph
      Nordic plus EU parliament expansion
      Real-time streaming intelligence
    ⚖️ Invariants Across All Horizons
      Public data only
      Political neutrality all parties equal
      GDPR Article 9 2 e and 2 g lawful bases
      Transparent methodology and uncertainty
      Hack23 ISMS ISO 27001 NIST CSF CIS

Reading guide. Horizon boundaries are capability gates, not hard dates — v2.0 features may ship continuously through 2026–2027 while v3.0 foundations (Bedrock evaluation, Knowledge-Base prototyping) begin in parallel. The four invariants are non-negotiable constraints that survive every architectural change.

---

2. 🟢 Horizon 1 — v1.x Baseline Capabilities

This is the accurate current state the future is built upon. It is intentionally detailed so the deltas in Horizons 2 and 3 are unambiguous. Cross-reference MINDMAP.md §1–§5 for the matching current-state surface.

mindmap
  root((🟢 v1.x Baseline<br/>Shipped Today))
    🌐 Delivery Surface
      Static HTML5 and CSS3 cyberpunk theme
      14 languages including RTL Arabic Hebrew
      WCAG 2.1 AA accessibility
      No JS frameworks for content
      Dashboards lazy-loaded TypeScript modules
        Chart.js and D3.js
        ~11 functional dashboards
        Party ministry anomaly seasonal pre-election politician
    🤖 Autonomous AI Newsroom
      14 agentic workflows gh-aw
      Claude Opus 4.8 generation
      Claude Sonnet 4.6 translation
      Zero human editors
      Daily articles in 14 languages
    📡 Data Foundation
      349 current MPs
      2494 historical politicians 1971 to 2024
      3.5M plus votes
      109000 plus documents
      Riksdagen API Regeringen g0v SCB IMF World Bank
      CIA platform JSON and CSV exports
    🧠 MCP Tooling
      riksdag-regering HTTP
      scb local
      world-bank local
      imf TypeScript client
      github filesystem memory sequential-thinking playwright
    🔍 Structured Tradecraft
      ACH competing hypotheses
      SWOT and PESTLE
      STRIDE political threat modeling
      Political-risk scoring
      OSINT and INTOP scorecards
    ☁️ Hosting and CI CD
      AWS CloudFront edge
      Multi-region S3 us-east-1 plus eu-west-1
      GitHub Pages disaster recovery
      GitHub Actions build and deploy

Baseline takeaway. v1.x already proves the hard parts: autonomous multilingual generation, public-data discipline, and structured analytic tradecraft. Horizon 2 raises analytic depth and party focus; Horizon 3 changes the delivery substrate from pre-rendered artifacts to managed serverless + AI services.

---

3. 🔵 Horizon 2 — v2.0 Static Deepening (2026–2027)

Strategic choice: keep the static architecture, go deeper. No migration in this horizon — every output remains a pre-rendered, CDN-served static artifact. Investment concentrates on party-focused analytics, OSINT/INTOP quality, multilingual richness, accessibility and performance.

mindmap
  root((🔵 v2.0 Static<br/>2026 to 2027))
    🏛️ Party-Focused Landscape Dashboards
      Party cohesion and discipline scoring
      Coalition dynamics and bloc alignment
      Party-vs-party comparison views
      Agenda and issue-salience tracking
      Government-vs-opposition divergence
      Rebel-vote and defection detection
    🕵️ Advanced OSINT and INTOP
      Network analysis co-sponsorship and committees
      Temporal pattern detection
      Geospatial valkrets patterns
      Anomaly detection on voting and attendance
      Source-graded evidence Admiralty scale
      INTOP scorecards per actor
    🌍 Richer Multilingual Products
      Expanded glossary political terminology
      Cultural-context adaptation per language
      Per-language structured data Schema.org
      Back-translation quality checks
    ♿ Accessibility and Performance
      WCAG 2.2 AA target
      Core Web Vitals all green
      Logical-property RTL refinement
      Progressive enhancement static-first
    🏗️ Architecture Stance
      Static HTML CSS retained
      AI confined to build and newsroom pipeline
      Outputs remain static artifacts
      CloudFront S3 plus Pages DR unchanged

v2.0 vs v3.0 distinction. In v2.0 the AI is a build-time author — it deepens analysis but the citizen still consumes a static page. In v3.0 the AI becomes a run-time interlocutor: citizens query a conversational assistant backed by Bedrock and Knowledge Bases, and developers consume a live API. v2.0 lowers analytical risk and cost while compounding domain depth that v3.0 later exposes interactively.

---

4. 🏗️ Horizon 3 — v3.0+ AWS Serverless (2028–2037)

Strategic choice: all-in AWS serverless. Full migration to managed compute and managed AI — no Kubernetes, no containers to operate. AWS Well-Architected aligned, multi-region resilient. The static corpus becomes a queryable, conversational, programmable intelligence platform.

mindmap
  root((🏗️ v3.0 plus Serverless<br/>2028 to 2037))
    🤖 Amazon Bedrock Intelligence
      Conversational political-intelligence assistant
      Bedrock Agents for multi-step analysis
      Foundation-model choice and routing
      Guardrails for neutrality and safety
    📚 Knowledge Bases and RAG
      RAG over 109000 plus documents
      Citation-grounded answers
      Vector index OpenSearch Serverless
      Freshness from daily ingestion
    🚪 API Gateway Public API Economy
      REST and GraphQL political-intelligence API
      Usage plans and throttling
      Free researcher tier
      Pro and enterprise tiers
    🔐 Cognito Identity and Personalization
      Sign-in for saved views and alerts
      API-consumer identity and keys
      Preference and topic subscriptions
      Privacy-preserving no political profiling
    📈 Predictive Analytics and Forecasting
      Seat-projection and coalition probability
      Vote-outcome prediction
      Monte Carlo with confidence intervals
      Historical backtesting transparency
    🔎 Semantic Search and Knowledge Graph
      Natural-language multilingual queries
      Neptune Serverless political graph
      Entity-relationship mining
      Graph-plus-vector hybrid retrieval
    🌍 Nordic and EU Expansion
      Folketing Storting Eduskunta
      EU Parliament 705 MEPs
      Cross-parliament comparison
      Federated multi-country corpus
    ⚡ Real-Time Streaming Intelligence
      Kinesis parliamentary feeds
      EventBridge event routing
      Step Functions orchestration
      Sub-second live updates
    🧱 Managed Serverless Substrate
      Lambda compute
      DynamoDB and Aurora Serverless v2
      Timestream time-series
      Zero infrastructure to manage

Horizon-3 grounding. Each branch names the concrete AWS service that realises it — Bedrock (assistant + agents + guardrails), OpenSearch Serverless (vectors), Neptune Serverless (graph), API Gateway (API economy), Cognito (identity), Kinesis/EventBridge/Step Functions (streaming), DynamoDB/Aurora Serverless v2/Timestream (state). Migration is incremental: read-paths move to serverless behind CloudFront before any write-path, preserving the v1.x DR posture during cutover.

---

5. 📡 Theme — Data & Sources

How the data surface widens across horizons: from Swedish open data + CIA exports (v1.x), to richer party-graded evidence (v2.0), to a federated Nordic/EU corpus with streaming ingestion (v3.0).

mindmap
  root((📡 Data and Sources<br/>Evolution))
    🟢 v1.x Foundation
      Riksdagen open data votes motions ledamoter
      Regeringen via g0v
      SCB Swedish ground truth
      IMF primary economic WEO FM IFS
      World Bank non-economic residue
      CIA platform JSON and CSV
    🔵 v2.0 Deepening
      Source-graded evidence Admiralty scale
      Cross-validation IMF vs SCB delta checks
      Committee-aligned spending GFS_COFOG
      Provenance metadata per claim
    🟣 v3.0 Federation and Streaming
      Folketing Storting Eduskunta open data
      EU Parliament open data 705 MEPs
      Kinesis real-time parliamentary feeds
      Knowledge-Base ingestion pipeline
      Vector and graph indexed corpus
    ⚖️ Provenance Discipline
      Economic claim cites IMF dataflow first
      Vintage annotation when older than 6 months
      Public-data-only no leaked or private data
      Primary-source URL or dok_id per claim

Source canon. The economic-data contract (ECONOMIC_DATA_CONTRACT.md v2.1) survives every horizon: IMF is primary for economics, World Bank is reserved for governance/environment/social residue, SCB is Swedish ground truth, and Riksdag open data is parliamentary primary. v3.0 adds federated Nordic/EU sources without weakening provenance rules.

---

6. 🕵️ Theme — OSINT / INTOP Tradecraft

Tradecraft maturity is the spine of the product. v1.x applies structured techniques in the newsroom; v2.0 productises them into dashboards and scorecards; v3.0 makes them conversational and predictive.

mindmap
  root((🕵️ OSINT and INTOP<br/>Tradecraft))
    🧩 Structured Techniques
      ACH competing hypotheses
      SWOT and PESTLE scanning
      Devil's advocacy and red team
      Attack-tree and STRIDE political variants
    🔗 Network and Pattern Analysis
      Co-sponsorship networks
      Committee-membership graphs
      Centrality bridging and clustering
      Temporal and geospatial patterns
    📑 Evidence and Sourcing
      Admiralty source grading
      Confidence scoring and uncertainty
      Citation-grounded claims
      INTOP scorecards per actor
    🟢 v1.x newsroom-applied
    🔵 v2.0 dashboard-productised
    🟣 v3.0 conversational and predictive
      Bedrock-assisted hypothesis testing
      Automated fact-checking with citations
      Predictive coalition and vote models

Ethics rail. Tradecraft is applied to public political behaviour only — votes, speeches, documents, declared affiliations. No surveillance, no private-life profiling, no psyops framing. The platform exposes accountability evidence; it never weaponises it.

---

7. 🏛️ Theme — Party & Coalition Analytics

The headline v2.0 investment. Party-focused analytics turn raw votes into landscape understanding: cohesion, bloc alignment, coalition viability and agenda movement.

mindmap
  root((🏛️ Party and Coalition<br/>Analytics))
    📊 Party Behaviour
      Cohesion and discipline scores
      Rebel-vote and defection detection
      Attendance and productivity
      Agenda and issue-salience tracking
    🤝 Coalition Dynamics
      Bloc alignment government vs opposition
      Coalition viability and stability
      Cross-party alignment matrices
      Alternative-coalition scenarios
    ⚖️ Comparison Views
      Party-vs-party head to head
      Historical drift 1971 to today
      Constituency valkrets breakdowns
    🟢 v1.x baseline dashboards
    🔵 v2.0 party-focused expansion
      Deeper cohesion analytics
      Coalition-math visualizations
    🟣 v3.0 predictive coalition intelligence
      Seat-projection models
      Coalition-probability forecasting
      Conversational what-if scenarios

Neutrality guarantee. Party analytics treat all eight Riksdag parties identically — same metrics, same methodology, same uncertainty disclosure. Comparison views are symmetric and never editorialise a "winner".

---

8. 🤖 Theme — AI / Bedrock Capabilities

AI moves from build-time author (v1.x/v2.0) to run-time intelligence (v3.0). Amazon Bedrock anchors the v3.0 conversational and RAG capabilities.

mindmap
  root((🤖 AI and Bedrock<br/>Capabilities))
    🟢 v1.x Build-Time AI
      Claude Opus 4.8 generation
      Claude Sonnet 4.6 translation
      gh-aw agentic workflows
      Outputs are static artifacts
    🔵 v2.0 Deeper Build-Time AI
      Higher-quality structured analysis
      Multilingual product enrichment
      Automated evidence grading
    🟣 v3.0 Run-Time Bedrock AI
      Conversational intelligence assistant
      Bedrock Agents multi-step analysis
      Knowledge Bases RAG citation-grounded
      Guardrails neutrality and safety
      Foundation-model routing and choice
    🔮 Forecasting and Verification
      Election and vote forecasting
      Real-time fact-checking
      Confidence-scored predictions
    ⚖️ AI Governance
      Hack23 AI Policy human-in-the-loop
      Explainable and transparent methods
      Bias detection and neutrality checks
      No user political profiling

Bedrock rationale. Managed foundation models with built-in Guardrails let Riksdagsmonitor add conversational and predictive intelligence without operating model infrastructure — consistent with the zero-infrastructure serverless strategy and the AI Policy's human-in-the-loop and transparency requirements.

---

9. 🏗️ Theme — Platform / AWS Serverless

The substrate change defining Horizon 3. Each capability maps to a named, managed AWS service; nothing requires servers, clusters or containers to operate.

mindmap
  root((🏗️ Platform and<br/>AWS Serverless))
    🧱 Compute and Orchestration
      Lambda functions
      Step Functions workflows
      EventBridge event bus
    🗄️ Data Stores
      DynamoDB key-value
      Aurora Serverless v2 relational
      Neptune Serverless graph
      OpenSearch Serverless vector and search
      Timestream time-series
    🚪 Edge and API
      API Gateway REST and GraphQL
      CloudFront retained at edge
      Cognito identity and authorization
    🤖 Managed AI
      Amazon Bedrock models and Agents
      Bedrock Knowledge Bases RAG
      Bedrock Guardrails
    ⚡ Streaming
      Kinesis data streams
      Real-time parliamentary feeds
    🟢 v1.x CloudFront S3 plus Pages DR
    🔵 v2.0 same static substrate
    🟣 v3.0 full serverless migration
      Zero infrastructure to manage
      AWS Well-Architected aligned
      Multi-region resilience

Migration discipline. The v1.x static site and GitHub Pages DR remain the resilience floor during cutover. Read paths migrate first behind CloudFront; identity, write paths and streaming follow once Cognito and observability are proven.

---

10. 🌍 Theme — Multilingual & Accessibility

A first-class, cross-horizon concern. v1.x ships 14 languages and WCAG 2.1 AA; v2.0 deepens quality and targets WCAG 2.2; v3.0 makes multilingual interactive via conversational and semantic search.

mindmap
  root((🌍 Multilingual and<br/>Accessibility))
    🗣️ Language Coverage
      14 languages today
      Germanic Romance Nordic Asian RTL
      English source Swedish primary
      Expansion targets EU and UN languages
    🔁 Translation Pipeline
      Version-controlled source strings
      Political terminology glossary
      LLM translation Sonnet 4.6
      Back-translation quality checks
    ♿ Accessibility
      WCAG 2.1 AA today
      WCAG 2.2 AA target v2.0
      RTL logical properties
      Screen-reader and keyboard support
    ⚡ Performance
      Core Web Vitals green
      Static-first progressive enhancement
      Lazy-loaded dashboard modules
    🟢 v1.x 14 languages static
    🔵 v2.0 richer products WCAG 2.2
    🟣 v3.0 multilingual conversational and semantic search

Accessibility as ethics. Multilingual + accessible delivery is part of the democratic mission — intelligence that citizens cannot read or perceive does not strengthen accountability. This theme is never deprioritised by the serverless migration.

---

11. ⚖️ Theme — Governance / Ethics / GDPR

The guardrail theme. These controls are invariant across all three horizons and gate every new capability.

mindmap
  root((⚖️ Governance Ethics<br/>and GDPR))
    🔐 Data Protection
      Public data only
      No hacked leaked or private data
      Data minimization and purpose limitation
      Storage limitation and integrity
    📜 GDPR Article 9
      Political opinion special category
      Lawful basis 9 2 e publicly made
      Lawful basis 9 2 g substantial public interest
      DPIA for high-risk processing
    🟦 Neutrality
      Equal treatment all parties
      Transparent methodology
      Documented uncertainty
      No psyops no propaganda
    🛡️ ISMS Alignment
      ISO 27001 2022
      NIST CSF 2.0
      CIS Controls v8.1
      NIS2 and EU CRA where applicable
    🤖 AI Governance
      Hack23 AI Policy
      Human-in-the-loop oversight
      Explainability and bias checks
      No user political profiling

Cross-link. This theme operationalises the Future Security Architecture and Threat Model, and inherits the Hack23 ISMS policies linked in the footer. v3.0 personalization (Cognito) is explicitly constrained: identity enables saved views and alerts, never political-profile inference.

---

12. 🌐 Theme — Ecosystem & Expansion

How Riksdagsmonitor grows outward — into Nordic and EU parliaments, a public API economy, and the wider Hack23 ecosystem.

mindmap
  root((🌐 Ecosystem and<br/>Expansion))
    🌍 Geographic Expansion
      Sweden today
      Nordic Folketing Storting Eduskunta
      EU Parliament 705 MEPs
      Cross-parliament comparative analysis
    🔌 API Economy
      API Gateway public API
      Free researcher tier
      Pro and enterprise tiers
      Knowledge-Base backed answers
    🤝 Hack23 Ecosystem
      Citizen Intelligence Agency engine
      EU Parliament Monitor
      European Parliament MCP
      Shared ISMS and tradecraft
    🎓 Stakeholders
      Citizens and journalists
      Researchers and think tanks
      Developers and integrators
      Civil-society accountability
    🟢 v1.x Sweden static
    🔵 v2.0 deeper Swedish coverage
    🟣 v3.0 federated multi-country and API

Expansion principle. Geographic and API expansion reuse the same tradecraft, neutrality and provenance rules. The European Parliament MCP and EU Parliament Monitor are existing Hack23 assets that de-risk the v3.0 cross-parliament leap.

---

13. 🤖 AI / LLM Capability-Evolution (2026–2037)

Major AI model upgrades are assumed annually, with competitor releases (OpenAI, Google, Meta, EU sovereign AI) evaluated at each cycle, and the architecture explicitly designed to accommodate paradigm shifts (quantum AI, neuromorphic computing). All adoption is governed by the Hack23 AI Policy with human-in-the-loop oversight.

13.1 AI Model Evolution Table (DevSecOps & development perspective)

Year	AI Model	DevSecOps Capability Evolution
2026	Opus 4.6–4.9	🟢 AI-assisted code review, automated test generation, agentic CI/CD workflows
2027	Opus 5.x	🔵 Predictive vulnerability detection, intelligent dependency management
2028	Opus 6.x	🟣 Multi-modal security analysis (code + architecture + runtime), automated threat modeling
2029	Opus 7.x	🟠 Autonomous security pipeline orchestration, self-healing build systems
2030	Opus 8.x	🔴 Near-expert automated security review, AI-driven architecture validation
2031–2033	Opus 9–10.x / Pre-AGI	⚪ Autonomous secure development lifecycle management
2034–2037	AGI / Post-AGI	⭐ Transformative software engineering with built-in security assurance

13.2 Translating the AI Curve into Product & Intelligence Unlocks

The same model generations unlock product, data and tradecraft capability — not only DevSecOps. The mindmap below maps each AI generation to the intelligence it enables.

mindmap
  root((🤖 AI Model Evolution<br/>to Capability Unlocks))
    2026 Opus 4.6 to 4.9 Green
      Build-time newsroom generation 14 languages
      AI-assisted code review and test generation
      Agentic gh-aw workflows
      Product unlock party-focused static dashboards
    2027 Opus 5.x Blue
      Predictive vulnerability detection
      Intelligent dependency management
      Intelligence unlock early forecasting prototypes
      Deeper OSINT evidence grading
    2028 Opus 6.x Purple
      Multi-modal analysis code architecture runtime
      Automated threat modeling
      Platform unlock Bedrock conversational assistant
      Knowledge Bases RAG go-live
    2029 Opus 7.x Orange
      Autonomous pipeline orchestration
      Self-healing build systems
      Intelligence unlock real-time fact-checking
      Nordic parliament onboarding
    2030 Opus 8.x Red
      Near-expert automated security review
      AI-driven architecture validation
      Intelligence unlock mature election forecasting
      Semantic search and knowledge graph maturity
    2031 to 2033 Opus 9 to 10 Pre-AGI White
      Autonomous secure development lifecycle
      Federated EU and Nordic corpus
      Predictive coalition intelligence at scale
    2034 to 2037 AGI Post-AGI Star
      Transformative engineering with built-in assurance
      Autonomous cross-parliament analysis
      Visionary global democratic-intelligence reach

13.3 Capability-Unlock Crosswalk

graph LR
    A[2026 Opus 4.6-4.9<br/>Build-time author]:::g --> B[2027 Opus 5.x<br/>Predictive prototypes]:::b
    B --> C[2028 Opus 6.x<br/>Bedrock run-time AI]:::p
    C --> D[2029 Opus 7.x<br/>Autonomous orchestration]:::o
    D --> E[2030 Opus 8.x<br/>Near-expert review]:::r
    E --> F[2031-2033 Pre-AGI<br/>Autonomous SDLC]:::w
    F --> G[2034-2037 AGI<br/>Transformative assurance]:::s
    classDef g fill:#4caf50,stroke:#2e7d32,color:#000000;
    classDef b fill:#42a5f5,stroke:#1565c0,color:#000000;
    classDef p fill:#ab47bc,stroke:#6a1b9a,color:#000000;
    classDef o fill:#ffa726,stroke:#e65100,color:#000000;
    classDef r fill:#ef5350,stroke:#b71c1c,color:#000000;
    classDef w fill:#eceff1,stroke:#546e7a,color:#000000;
    classDef s fill:#ffd54f,stroke:#f57f17,color:#000000;

Governance note. Each annual model adoption is a Change Management event: competitor benchmarking, neutrality/bias evaluation, and human-in-the-loop sign-off precede production use. Paradigm-shift contingencies (quantum/neuromorphic) are tracked as architecture-accommodation requirements, not committed deliverables.

---

🛰️ Theme — Political-Intelligence Capability Catalog (to 2037) — the master OSINT/INTOP map

Why this section exists. Sections 6–9 describe tradecraft, party analytics and AI by theme. This section is the single authoritative capability catalog an intelligence operative would demand: it names every OSINT / INTOP / political-intelligence capability the platform must field by 2037, organises them along the classic intelligence cycle (Direction → Collection → Processing → Analysis → Production → Dissemination → Feedback), maps each to a horizon, names the AI method that unlocks it, and states the operational benefit (why it matters for democratic accountability). It is the master that the sibling FUTURE_ARCHITECTURE.md, FUTURE_WORKFLOWS.md, FUTURE_DATA_MODEL.md, FUTURE_FLOWCHART.md, FUTURE_STATEDIAGRAM.md, FUTURE_SWOT.md, FUTURE_THREAT_MODEL.md and FUTURE_SECURITY_ARCHITECTURE.md each realise in their own idiom.

Gap honesty — what today's stack does not yet do. The v1.x newsroom already fields ~24 analysis artifacts (F3EAD, ICD 203, ACH, Admiralty grading, PIR roll-forward, calibrated WEP forecasting, DISARM media-framing). That is a strong single-source, build-time, document-centric baseline. The capabilities below are the deliberate gaps an operative would flag: multi-INT fusion (we read documents, not the money/lobbying/social context around them), entity resolution across registries, indications-and-warning tripwires, agent-based wargaming, continuous calibration, foreign-interference (FIMI) early-warning, content provenance / deepfake defence, and runtime estimative products. Closing them is the visionary spine of v2.0 → v3.0+.

PI.0 The intelligence cycle as the platform's operating loop

graph LR
    DIR["🎯 Direction<br/>PIRs and tripwires"]:::g --> COL["📡 Collection<br/>Multi-INT fusion"]:::b
    COL --> PROC["⚙️ Processing<br/>Extraction and provenance"]:::p
    PROC --> ANA["🧠 Analysis<br/>SAT + forecasting + I&amp;W"]:::o
    ANA --> PROD["📑 Production<br/>Estimative products"]:::r
    PROD --> DIS["📣 Dissemination<br/>COP and briefings"]:::w
    DIS --> FB["🔁 Feedback<br/>Calibration and PIR roll-forward"]:::s
    FB --> DIR
    classDef g fill:#4caf50,stroke:#2e7d32,color:#000000;
    classDef b fill:#42a5f5,stroke:#1565c0,color:#000000;
    classDef p fill:#ab47bc,stroke:#6a1b9a,color:#ffffff;
    classDef o fill:#ffa726,stroke:#e65100,color:#000000;
    classDef r fill:#ef5350,stroke:#b71c1c,color:#ffffff;
    classDef w fill:#26c6da,stroke:#00838f,color:#000000;
    classDef s fill:#ffd54f,stroke:#f57f17,color:#000000;

Ethics rail (binds every capability below). All capabilities operate on public political behaviour only — votes, speeches, documents, declared interests, public financial disclosures, public discourse. No private-life surveillance, no citizen profiling, no psyops. Political opinions are GDPR Article 9 special-category data (lawful bases 9(2)(e) manifestly-public / 9(2)(g) substantial-public-interest). Every externally visible judgment is human-signed per the Hack23 AI Policy.

PI.1 Pillar — Collection & Multi-INT Fusion 📡 (close the single-source gap)

mindmap
  root((📡 Collection and<br/>Multi-INT Fusion))
    🔵 v2.0 Build-Time Fusion
      Cross-source entity resolution registries
      Lobbying and revolving-door mapping
      Declared-interest versus vote conflict screening
      Party-funding and campaign-finance tracing
      Procurement and public-spend anomaly screening
    🟣 v3.0 Runtime Fusion
      OSINT plus FININT plus GEOINT plus ECONINT mesh
      Privacy-bounded SOCMINT discourse listening
      EU transparency register cross-walk
      Court and agency decision ingestion
    🌟 2032 to 2037 Always-On Fusion
      Continuous multi-parliament collection mesh
      Tip-and-cue automated re-tasking
      Federated Nordic and EU evidence pool
    ⚖️ Collection Rails
      Public sources only no scraping of private data
      Provenance tag on every collected item
      Admiralty source grading at intake

Operative benefit. Documents alone show what an MP did; fusion shows why it may matter — the donor, the prior employer, the lobby meeting, the constituency money flow. Triangulation across independent source families is the single biggest jump in judgment confidence and the strongest defence against being spun by any one source.

PI.2 Pillar — Processing & Provenance ⚙️ (make every claim verifiable and tamper-evident)

mindmap
  root((⚙️ Processing and<br/>Provenance))
    🔵 v2.0 Build-Time
      Multilingual entity event relation extraction
      Temporal knowledge-graph construction
      Source-graded evidence ledger Admiralty
      Citation-grounded claim binding dok_id
    🟣 v3.0 Runtime
      Content provenance and C2PA credentials
      Deepfake and synthetic-media detection
      Cross-language coreference resolution
      Streaming event extraction from live feeds
    🌟 2032 to 2037
      Self-verifying provenance chain end to end
      Cryptographic evidence chain-of-custody
    ⚖️ Processing Rails
      No claim without a primary-source anchor
      Reproducible extraction same input same output
      Vintage discipline on every dated fact

Operative benefit. In an era of synthetic media and information laundering, provenance is the product. A judgment is only as trustworthy as its weakest evidence link; tamper-evident chain-of-custody and deepfake screening let the platform refuse to cite what it cannot verify — and prove it later.

PI.3 Pillar — Analysis & Tradecraft AI 🧠 (the "much more intelligence" the brief asks for)

mindmap
  root((🧠 Analysis and<br/>Tradecraft AI))
    🔵 v2.0 Build-Time Tradecraft
      Structured Analytic Techniques at scale
      ACH with Bayesian hypothesis updating
      Argument-mining claim evidence warrant
      Narrative and framing lifecycle tracking
      Stance and issue-position drift per actor
      Influence and power-broker network analytics
      Anomaly explanation not just detection
    🟣 v3.0 Runtime Analysis
      Calibrated probabilistic forecasting engine
      Indications and Warning tripwire system
      Causal inference policy-impact modelling
      Agent-based legislative simulation wargaming
      FIMI and coordinated-inauthentic-behaviour detection
      Leadership analysis public statements only
    🌟 2032 to 2037 Autonomous Analysis
      Generative scenario and wildcard synthesis
      Graph-neural influence prediction
      Neuro-symbolic explainable judgments
      Continuous self-calibrating forecasts
    ⚖️ Analysis Rails
      ICD 203 analytic integrity on every product
      Party-symmetry neutrality arithmetic
      Documented uncertainty and confidence bands
      Red-team and devils-advocate built in

Operative benefit. This pillar converts evidence into decision-relevant judgment. Indications-and-Warning tripwires turn the platform from a rear-view mirror into an early-warning radar (government collapse, snap election, coalition rupture, budget crisis, integrity scandal). Calibrated forecasting plus continuous Brier scoring means the platform earns trust by being measurably right over time, not merely confident.

PI.4 Pillar — Production & Dissemination 📣 (turn analysis into products people can act on)

mindmap
  root((📣 Production and<br/>Dissemination))
    🔵 v2.0 Static Products
      Neutral actor and party scorecards
      Evidence-graded static intelligence briefs
      Coalition-math and landscape dashboards
    🟣 v3.0 Runtime Products
      Common Operating Picture live dashboard
      Estimative product NIE-style key judgments
      Autonomous Riksdag Daily Brief
      Personalized PIR-driven briefings
      Tip-and-cue watchlist alerting
      Conversational intelligence analyst agentic RAG
      Counter-disinformation rapid-response briefs
    🌟 2032 to 2037 Decision Support
      Election-night live intelligence cell
      Real-time democracy-health index
      Cross-parliament comparative COP
    ⚖️ Dissemination Rails
      Human signs every published judgment
      Confidence and source grade shown to reader
      Equal treatment of all parties in every product
      No personalized political targeting of citizens

Operative benefit. Raw analysis that nobody reads at the right moment is wasted. A Common Operating Picture and tip-and-cue alerting put the right judgment in front of the right citizen, journalist or researcher at the decision-relevant moment — while neutrality rails and visible confidence grades keep the product a transparency tool, never a persuasion weapon.

PI.5 Pillar — Direction, Feedback & Assurance ⚖️ (steer the loop and prove integrity)

mindmap
  root((⚖️ Direction Feedback<br/>and Assurance))
    🔵 v2.0
      PIR generation and roll-forward
      Calibration ledger automation Brier scoring
      Neutrality and bias-symmetry auditing
      Reproducibility and full audit trail
    🟣 v3.0
      Continuous I&amp;W tripwire re-tasking
      Adversarial red-team of the pipeline itself
      Counter-AI prompt-injection and poisoning defence
      Analytic-integrity ICD 203 and 206 scoring
    🌟 2032 to 2037
      Self-auditing assurance-by-construction
      Federated calibration across parliaments
    ⚖️ Assurance Rails
      Every inference human-accountable
      Kill-switch and least-privilege agent scoping
      No autonomous partisan or psyops action

Operative benefit. Direction and feedback are what separate an intelligence organisation from a content firehose. Auto-generated PIRs focus collection on what actually matters; continuous calibration and adversarial self-testing keep the platform honest; counter-AI defences protect the integrity of judgments against actors who would poison or prompt-inject the analytic pipeline.

PI.6 Capability → Horizon → AI-Method → Benefit crosswalk (the operative's master table)

Legend: 🔵 v2.0 static-deep (2026–2027) · 🟣 v3.0 serverless runtime (2028–2031) · 🌟 autonomous era (2032–2037). Every row is a target, governed by the AI Policy, on public data, neutral across parties.

#	Capability	Pillar	Horizon	AI method / unlock	Operative benefit (why)
C1	Cross-source entity resolution	Collection	🔵	Record-linkage + embedding similarity	One actor → full footprint across registries
C2	Lobbying & revolving-door mapping	Collection	🔵	Graph extraction + temporal joins	Exposes hidden influence channels
C3	Declared-interest vs vote conflict screen	Collection	🔵	Rule + LLM reconciliation	Surfaces potential conflicts, neutrally
C4	Party-funding / campaign-finance tracing	Collection	🔵→🟣	Financial-document IE	Follow-the-money accountability
C5	Procurement / public-spend anomaly screen	Collection	🔵→🟣	Outlier detection + XAI	Corruption indicators on agencies
C6	Multi-INT fusion mesh (OSINT+FININT+GEOINT+ECONINT)	Collection	🟣	Multi-source fusion + Neptune graph	Triangulated high-confidence judgments
C7	Privacy-bounded SOCMINT listening	Collection	🟣	Aggregate stance/salience models	Links elite behaviour to public salience
C8	Content provenance & C2PA credentials	Processing	🟣	Provenance signing + verification	Tamper-evident evidence
C9	Deepfake / synthetic-media detection	Processing	🟣	Multimodal forensics models	Refuse to cite what cannot be verified
C10	Multilingual entity/event/relation extraction	Processing	🔵	Transformer IE over 14 langs	Machine-readable knowledge graph
C11	Temporal knowledge-graph reasoning	Processing→Analysis	🟣	Neptune + GNN	Reason over evolving networks
C12	Structured-Analytic-Technique automation	Analysis	🔵	LLM-orchestrated ACH/KAC/Premortem	ICD 203-grade rigour at scale
C13	Calibrated probabilistic forecasting engine	Analysis	🟣	Ensemble + Brier calibration	Measurably-right, trust-earning forecasts
C14	Indications & Warning tripwire system	Analysis	🟣	Indicator models + thresholds	Early-warning radar, not rear-view mirror
C15	Causal policy-impact inference	Analysis	🟣	Diff-in-diff / synthetic control	Did the law actually change outcomes
C16	Agent-based legislative wargaming	Analysis	🟣→🌟	Multi-agent simulation	Stress-test coalition/vote scenarios
C17	Argument-mining & rhetoric analysis	Analysis	🔵	Claim-evidence-warrant extraction	Detect spin and logical fallacy
C18	Narrative / framing lifecycle tracking	Analysis	🔵	DISARM TTP + topic dynamics	See how frames originate and mutate
C19	Influence / power-broker network analytics	Analysis	🔵→🟣	Centrality + diffusion + GNN	Identify kingmakers and brokers
C20	FIMI / coordinated-inauthentic-behaviour detection	Analysis	🟣	Network + behavioural anomaly	Foreign-interference early-warning
C21	Common Operating Picture (live)	Dissemination	🟣	Real-time fusion + streaming	Right judgment at decision moment
C22	Estimative products (NIE-style key judgments)	Production	🟣	RAG + SAT synthesis	Confidence-scored standing estimates
C23	Autonomous Riksdag Daily Brief	Production	🟣	Bedrock agents + RAG	PDB-style daily situational product
C24	Tip-and-cue watchlist alerting	Dissemination	🟣	I&W + EventBridge	Never miss a tracked actor/issue move
C25	Conversational intelligence analyst	Dissemination	🟣	Agentic RAG over corpus	Ask-anything grounded answers
C26	Election-night / crisis live cell	Dissemination	🌟	Streaming + live forecasting	High-tempo real-time intelligence
C27	Real-time democracy-health index	Production	🌟	Composite indicator model	Single neutral accountability barometer
C28	PIR auto-generation & roll-forward at scale	Direction	🔵→🟣	LLM requirement synthesis	Focus collection on what matters
C29	Continuous calibration & self-improvement	Feedback	🟣→🌟	Online Brier feedback loop	Forecasts improve provably over time
C30	Neutrality / bias-symmetry auditing AI	Assurance	🔵	Per-party arithmetic verification	Structural impartiality, not hope
C31	Adversarial red-team of the pipeline	Assurance	🟣	Automated attack generation	Find blind spots before adversaries
C32	Counter-AI integrity defence	Assurance	🟣	Prompt-injection / poisoning guards	Protect judgments from manipulation

PI.7 Maturity arc — where each pillar lives by year

timeline
    title Political-Intelligence Capability Maturity (2026 to 2037)
    2026 to 2027 v2.0 Static-Deep : Entity resolution and conflict screening : SAT automation and argument mining : Narrative tracking and influence networks : Neutrality auditing and PIR roll-forward
    2028 to 2029 v3.0 Runtime : Multi-INT fusion mesh and provenance : Calibrated forecasting and I and W tripwires : Conversational analyst and Daily Brief : FIMI detection and counter-AI defence
    2030 to 2031 v3.0 Mature : Causal policy-impact inference : Agent-based wargaming : Common Operating Picture : Estimative NIE-style products
    2032 to 2037 Autonomous : Always-on multi-parliament fusion : Generative scenario synthesis : Election-night live cell : Real-time democracy-health index

The single-sentence vision. By 2037 Riksdagsmonitor evolves from an AI newsroom that explains yesterday into a governed, neutral, public-source political-intelligence service that warns about tomorrow — fusing many open-source intelligence families, reasoning with calibrated tradecraft, and proving its own integrity on every judgment, for any citizen, in 14+ languages.

---

🦾 Theme — Agentic AI & Autonomous Multi-Agent Operations (what future AI agents unlock)

The single biggest visionary lever after v2.0+ is not a bigger model — it is agency: many specialised, governed AI agents that plan, use tools (MCP), verify each other, and run the platform under human-in-the-loop oversight rather than under a single person's manual workload. This theme answers the question "what is really possible with future AI agents?" and is the direct mitigation for the single-developer bus factor (SWOT W1) and manual content load (W5), while amplifying AI-enhanced analytics (O7). Today's 14 gh-aw build-time workflows are the seed; the maps below show the autonomous, multi-agent organisation they grow into.

mindmap
  root((🦾 Agentic AI<br/>Autonomous Operations))
    🟢 v1.x Seed Agents Today
      14 gh-aw newsroom workflows
      Multi-pass analysis aggregate render PR
      Claude Opus 4.8 plus Sonnet 4.6
      Human merges every pull request
    🔵 v2.0 Coordinated Build-Time Crew
      Editor agent commissions and outlines
      Researcher agent pulls MCP evidence
      Fact-checker agent verifies dok_id and votes
      Neutrality-auditor agent flags partisan drift
      Translator agent 14 languages back-translation
      Reviewer agent gates before human merge
    🟣 v3.0 Run-Time Bedrock Agents
      Bedrock Agents multi-step tool use
      Agentic RAG over 109000 documents
      Agent-to-agent A2A handoff protocol
      MCP tool ecosystem as agent skills
      Always-on chamber-watch streaming agent
      Self-verifying citation loop
    🌟 2031 to 2037 Autonomous Org
      Agent swarm cross-parliament coverage
      Self-healing pipelines and dependency agents
      Continuous assurance-by-construction
      Human sets intent agents execute and prove
    ⚖️ Agent Governance Rails
      Human-in-the-loop sign-off on every write
      Bedrock Guardrails neutrality and safety
      Full audit trail and reproducibility
      Kill-switch and least-privilege scoping
      No autonomous partisan or psyops action

The multi-agent newsroom (v2.0 → v3.0). The leap is from one generalist workflow to a division of labour of cooperating specialist agents, each with a narrow remit and an explicit verification duty toward the others. A neutrality-auditor agent and a fact-checker agent are first-class members — not afterthoughts — so quality and impartiality are enforced structurally, not by hope.

mindmap
  root((🤖 Future Agent Roles<br/>Closing W1 and W5))
    🛠️ Platform-Sustaining Agents W1
      Dependency-update and patch agent
      Security-triage and CodeQL-fix agent
      Test-generation and coverage agent
      Refactor and tech-debt agent
      Docs-sync agent keeps portfolio current
      Bus-factor goal repo runs without one person
    📝 Content and Intelligence Agents W5 O7
      Multilingual newsroom crew 14 languages
      Evidence-grading and source-Admiralty agent
      Forecasting and scenario-tree agent
      Devil's-advocate and red-team agent
      Anomaly-watch agent on votes and attendance
    📣 Growth and Outreach Agents W3
      SEO and discoverability agent
      Social-syndication draft agent human-approved
      Partnership-lead research agent O4
      Civic-education lesson drafting agent O5
      Business-policy alert drafting agent O6
    🔒 Trust and Assurance Agents
      Neutrality-auditor agent all parties equal
      Privacy and GDPR Article 9 guard agent
      Provenance and vintage-discipline agent
      Reproducibility and audit-log agent

Why this is realistic, not hype. Every agent above is a governed automation of work the project already does by hand — dependency bumps, translation, evidence grading, SEO, partnership outreach. Future model generations (see §13) raise reliability and tool-use depth; the Hack23 AI Policy keeps a human accountable for every externally-visible action. Outreach and social agents draft; a human publishes. No agent ever takes a partisan action, profiles a citizen, or publishes unverified claims — the four invariants bind agents exactly as they bind humans.

---

🌱 Theme — Sustainability, Growth & Audience Expansion

A future map that only deepens analytics but ignores sustainability and reach leaves SWOT W2 (no revenue), W3 (limited marketing), O4 (research partnerships), O5 (civic-education market) and O6 (business intelligence) unaddressed. This theme gives each a concrete home, sequenced low-risk-first: audience and credibility on the static substrate (v2.0), monetisation and segment products on the serverless substrate (v3.0).

mindmap
  root((🌱 Sustainability Growth<br/>and Audience))
    💸 Revenue and Sustainability W2
      v2.0 grants and co-funding Vinnova EU
      v3.0 freemium API researcher tier
      v3.0 developer pro and enterprise tiers
      Break-even target ARR not yet achieved
      Funds contributor hire reduces W1
    📣 Marketing and Reach W3
      Content marketing daily AI articles as funnel
      SEO structured data and RSS syndication
      Newsroom output is the marketing engine
      Social and media outreach human-approved
      Open-source visibility and OpenSSF badges
    🎓 Research Partnerships O4
      University and think-tank collaborations
      Bulk export and reproducible datasets
      Co-publication and citation-ready evidence
      Institutional data subscriptions target
    🏫 Civic-Education Market O5
      Reading-level-adapted explainers
      Teacher resources and lesson modules
      How-Parliament-works interactive guides
      NGO and school distribution partners
    🏢 Business Intelligence O6
      Industry policy-monitoring dashboards
      Legislative-change alerting via API
      Stakeholder and committee mapping
      Regulatory early-warning reports
    🤝 Community and Bus-Factor W1
      Contributor onboarding and good-first-issues
      Comprehensive docs portfolio
      Agent-assisted maintenance see Agentic theme
      Foundation-model governance long-term

Sequencing logic. Audience, credibility and grant funding are built on the low-cost static substrate in v2.0 — the autonomous newsroom is the marketing flywheel (every cited, neutral, multilingual article is discoverable SEO surface and a reason to cite the platform). Paid segment products (research, civic-education, business-intelligence) land in v3.0 once the API economy, Cognito identity and metering exist. Crucially, monetisation never gates citizen access: the public static tier and 14-language newsroom stay free forever; paid tiers sell scale, automation and bespoke segments, not democratic transparency itself.

Neutrality & ethics rail. Growth must not compromise impartiality. Business-intelligence and outreach products use the same symmetric, source-cited methodology as the public site; the platform sells access and tooling, never favourable framing. No client — commercial, academic or political — can purchase editorial influence, and no audience-growth tactic may rely on partisan targeting.

---

✅ SWOT → Future Coverage Crosswalk

Final-review traceability. This table closes the loop on the originating request — "Is every opportunity, weakness and improvement area in SWOT.md addressed in the future architectures & mindmaps?" Every one of the 27 SWOT.md items maps to a concrete future home (this mindmap, FUTURE_SWOT.md, FUTURE_ARCHITECTURE.md) and the horizon that delivers it. All targets remain targets, never claimed as achieved.

SWOT item	Future home (this doc / portfolio)	Horizon
S1 50+ yr data	§5 Data & Sources, §16 IMF surface	leveraged all
S2 14 languages	§10 Multilingual & Accessibility	leveraged all
S3 CIA integration	§5 Data, §12 Ecosystem	leveraged all
S4 Static architecture	§3 v2.0, §9 Platform	v1.x→v3.0
S5 Transparent ISMS	§11 Governance	leveraged all
S6 riksdag-regering-mcp	§5 Data, §8 AI	leveraged all
S7 Agent ecosystem	§14 Agentic AI	v2.0→vision
S8 Visualizations	§3 v2.0, §7 Party Analytics	v2.0→v3.0
S9 Tradecraft methodology	§6 OSINT/INTOP	leveraged all
S10 Economic data fusion	§5 Data, §16 IMF surface	leveraged all
W1 Single-developer	§14 Agentic AI (platform-sustaining agents), §15 community	v2.0→vision
W2 No revenue	§15 Revenue, §12 API economy	v2.0→v3.0
W3 Limited marketing	§15 Marketing & Reach	v2.0
W4 Static limitations	§4 v3.0 Serverless, §9 Platform	v3.0
W5 Manual content	§14 Agentic AI (content crew)	v2.0→v3.0
O1 Nordic expansion	§12 Ecosystem, §4 v3.0	v3.0
O2 EU Parliament	§12 Ecosystem, §4 v3.0	v3.0
O3 API monetization	§12 API economy, §15 Revenue	v3.0
O4 Research partnerships	§15 Research Partnerships	v2.0→v3.0
O5 Civic-education market	§15 Civic-Education	v2.0→v3.0
O6 Business intelligence	§15 Business Intelligence	v3.0
O7 AI predictive analytics	§8 AI/Bedrock, §14 Agentic AI	v3.0→vision
T1 Competing platforms	§7 Party Analytics, §15 differentiation	all
T2 EU CRA burden	§11 Governance (ISMS mapping)	all
T3 Riksdag API changes	§5 Data (provenance & fallback)	all
T4 Market consolidation	§15 niche focus, §11 Governance	all
T5 Budget sustainability	§15 Revenue & community, ⚠️ Roadmap Risk	all

Coverage verdict. All 10 strengths are explicitly leveraged, all 5 weaknesses have a named mitigation horizon, all 7 opportunities have a delivery theme, and all 5 threats have a mitigating stance — with the previously-thin items (W1, W3, O4, O5, O6) now given first-class mindmap treatment in §14–§15. The deeper strategic numbers (pricing, ARR targets, stakeholder power–interest, competitive landscape) live in FUTURE_SWOT.md; the architectural realisation lives in FUTURE_ARCHITECTURE.md.

---

🎯 Stakeholder Value by Horizon

Capability only matters if it reaches stakeholders. This map shows what each audience gains as the platform crosses horizons — keeping the democratic-accountability mission central.

mindmap
  root((🎯 Stakeholder Value<br/>by Horizon))
    🏛️ Citizens and Voters
      v1.x readable static intelligence 14 languages
      v2.0 clearer party and coalition picture
      v3.0 ask-anything conversational assistant
    📰 Journalists
      v1.x daily AI articles to cite
      v2.0 source-graded evidence and scorecards
      v3.0 real-time fact-checking and API queries
    🎓 Researchers and Think Tanks
      v1.x open historical dataset 1971 to 2024
      v2.0 network and pattern analytics
      v3.0 knowledge-graph and bulk API access
    👨‍💻 Developers and Integrators
      v1.x open-source repo and MCP servers
      v2.0 richer static data artifacts
      v3.0 API Gateway tiers and Cognito keys
    🤝 Civil Society and Accountability
      v1.x transparent neutral methodology
      v2.0 deeper oversight of party behaviour
      v3.0 cross-parliament comparative accountability

⚠️ Roadmap Risk & Uncertainty

A future map must surface its own risks. These are the dominant uncertainties on the path to v3.0, each paired with a mitigating stance already embedded in the roadmap.

mindmap
  root((⚠️ Roadmap Risk<br/>and Mitigation))
    🧱 Migration Risk
      Static-to-serverless cutover complexity
      Mitigation read-paths first then writes
      Mitigation Pages DR retained as floor
    💸 Cost and Sustainability
      Bedrock and serverless run-time cost
      Mitigation usage tiers and caching
      Mitigation v2.0 stays low-cost static
    🤖 AI Reliability
      Hallucination and bias risk
      Mitigation RAG citation-grounding
      Mitigation Guardrails and human-in-the-loop
    📜 Regulatory
      GDPR Article 9 and NIS2 and CRA
      Mitigation public-data-only invariant
      Mitigation DPIA and continuous compliance
    ⚖️ Neutrality and Trust
      Perceived partisanship risk
      Mitigation symmetric metrics all parties
      Mitigation transparent methodology and uncertainty
    🌍 Expansion Complexity
      Multi-country data heterogeneity
      Mitigation reuse EU Parliament MCP assets
      Mitigation provenance discipline preserved

ISMS Control Mapping (roadmap → controls)

Roadmap element	ISO 27001:2022	NIST CSF 2.0	CIS v8.1	Guardrail
Serverless migration (v3.0)	A.8.25 Secure development	PR.PS	CIS 16	Read-path-first cutover, Pages DR floor
Bedrock / RAG assistant	A.5.23 Cloud services	GV.SC / ID.RA	CIS 15	Guardrails + human-in-the-loop (AI Policy)
Cognito personalization	A.5.15 Access control	PR.AA	CIS 5, 6	No political profiling, least privilege
API Gateway economy	A.8.26 App security req.	PR.PS / DE.CM	CIS 16	Usage plans, throttling, key scoping
Streaming ingestion	A.8.16 Monitoring	DE.CM / DE.AE	CIS 8	Provenance + vintage discipline
Cross-parliament data	A.5.34 Privacy & PII	GV.OC	CIS 3	Public-data-only, GDPR Art. 9 bases

Risk posture. The roadmap is deliberately staged so analytical value (v2.0) lands on the low-risk static substrate before the higher-risk serverless leap (v3.0). Every new capability inherits the same four invariants, and security/privacy controls are mapped continuously per the Future Security Architecture and Threat Model.

---

📋 Capability Matrix (horizon-aligned)

Future values are targets, not current achievements. Current-state values mirror MINDMAP.md and ARCHITECTURE.md.

Capability	🟢 v1.x Current	🔵 v2.0 Target 2026–2027	🟣 v3.0 Target 2028–2030	🌟 2031–2037 Vision
Architecture	Static HTML/CSS + CDN	Static, deepened	AWS serverless migration	Fully serverless multi-region
Delivery	Pre-rendered pages	Pre-rendered, richer	Conversational + API + pages	Real-time streaming intelligence
AI Role	Build-time author	Deeper build-time author	Run-time Bedrock assistant	Autonomous, assurance-built-in
Agent Autonomy	14 single-pass workflows	Coordinated build-time crew	Run-time Bedrock multi-agent	Self-healing autonomous org
AI Models	Opus 4.8 generation	Opus 5.x	Opus 6.x–8.x	Pre-AGI → AGI
Party Analytics	Baseline dashboards	Party-focused expansion	Predictive coalition models	Cross-parliament predictive
Search	Keyword/static	Faceted + graded	Semantic + knowledge graph	Conversational omnisearch
Languages	14	14, richer quality	14 + EU expansion	EU + UN languages target
Parliaments	1 (Sweden)	1 (deeper)	+ Nordic + EU	Federated multi-country
Data Latency	Daily batch	Daily, validated	Near-real-time	Sub-second streaming
API	None	None (static)	API Gateway public API	API economy with tiers
Identity	None	None	Cognito personalization	Identity-aware federation
Revenue	None (volunteer)	Grants / co-funding	Freemium + tiered API	Sustainable API economy
Audience / Reach	Organic + newsroom	SEO + partnerships + civic-ed	Segment products (research/BI)	Cross-parliament reach
Compliance	ISO/NIST/CIS aligned	+ WCAG 2.2 target	+ NIS2 / CRA attestation	Continuous compliance
Ethics Rails	Public-data, neutral	Unchanged invariant	Unchanged invariant	Unchanged invariant

---

🌐 Evolving the Current IMF Mindmap toward the Future Data-Sources Surface

Baseline: the already-implemented IMF subtree is documented in MINDMAP.md §IMF. The mindmap below extends that baseline with future capabilities (commercial-provider redundancy, websocket feeds, cross-validation worker) without removing today's pure-TS client foundation.

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

mindmap
  root((Riksdagsmonitor Data Sources))
    IMF PRIMARY ECONOMIC
      WEO
        NGDP_RPCH real GDP growth T plus 5 projections
        PCPIPCH CPI inflation
        LUR unemployment rate
        GGXWDG_NGDP general government debt percent GDP
        BCA_NGDPD current account percent GDP
      Fiscal Monitor FM
        Cyclically-adjusted balance
        Primary balance
        EDP-consistent debt
      IFS
        Monthly CPI
        Policy rates
        Reserves
      BOP
        Balance of payments quarterly
      DOTS
        Bilateral trade flows monthly
      GFS_COFOG
        02 Defence FoU
        07 Health SoU
        09 Education UbU
        10 Social protection SfU
      PCPS
        Commodity benchmarks
      ER
        SEK exchange rates daily
      MFS_IR and MFS_PR
        Monetary survey
    World Bank NON-ECONOMIC RESIDUE
      WGI Governance CC RL VA GE RQ PV
      Environment CO2 renewables forest
      Social residue education participation
      Defence depth MS MIL historicals
    SCB SWEDISH GROUND TRUTH
      AKU monthly labour
      KPI monthly inflation
      Regional and municipal
      Budget execution
    Riksdag Open Data PARLIAMENTARY PRIMARY
      Documents mot prop bet skr
      Voteringar
      Anforanden
      Ledamoter
    Future v3.0 Additions
      Nordic and EU open data federation
      Kinesis streaming ingestion
      Knowledge-Base RAG index
      Cross-validation worker IMF vs SCB

Canonical rule. Every economic claim in a Riksdagsmonitor article cites an IMF dataflow first; World Bank citations are reserved for governance, environment and social residue (the classes IMF does not publish). SCB is the Swedish-specific ground truth layer. See ECONOMIC_DATA_CONTRACT.md v2.1 for the banned-phrase list and vintage discipline (>6 mo → annotation). v3.0 federation and streaming extend, but never relax, this provenance discipline.

---

📚 Related Documents

Riksdagsmonitor Architecture Portfolio

Document	Focus	Description
🏛️ Architecture	🏗️ C4 Models	System context, containers, components
📊 Data Model	📊 Data	Entity relationships and data dictionary
🔄 Flowchart	🔄 Processes	Business and data flow diagrams
📈 State Diagram	📈 States	System state transitions and lifecycles
🧠 Mindmap	🧠 Concepts	Current system conceptual relationships
🧠 Future Mindmap	🔮 Concepts	Capability expansion plans (this document)
💼 SWOT	💼 Strategy	Strategic analysis and positioning
💼 Future SWOT	🔮 Strategy	Future strategic opportunities
🛡️ Security Architecture	🔒 Security	Current security controls and design
🚀 Future Security	🔮 Security	Planned security improvements
🎯 Threat Model	🎯 Threats	STRIDE/MITRE ATT&CK analysis
🚀 Future Architecture	🔮 Evolution	Architectural evolution roadmap
📊 Future Data Model	🔮 Data	Enhanced data architecture plans
🔄 Future Flowchart	🔮 Processes	Improved process workflows
📈 Future State Diagram	🔮 States	Advanced state management

Hack23 ISMS Policies

• 🛡️ Information Security Policy — Top-level governance
• 🧪 Secure Development Policy — Architecture documentation requirements
• 🤖 AI Policy — AI usage and human-in-the-loop governance
• 🔄 Change Management — Model-adoption and migration change control
• 🏷️ Classification Framework — CIA triad classification

---

📋 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

🌐 Platforms	📦 Open-Source Projects	🛡️ Governance & Standards
🗳️ Riksdagsmonitor — Swedish Parliament intelligence 🇪🇺 EU Parliament Monitor — European coverage 🕵️ Citizen Intelligence Agency — political-data engine 🌐 Hack23 AB — corporate site 📰 Hack23 Blog — engineering & policy 💼 Hack23 on LinkedIn	🗳️ Hack23/riksdagsmonitor 🕵️ Hack23/cia 🇪🇺 Hack23/euparliamentmonitor 🔌 Hack23/european-parliament-mcp ✅ Hack23/cia-compliance-manager 🥋 Hack23/black-trigram 🏠 Hack23/homepage	🛡️ Hack23 ISMS-PUBLIC — public ISMS 🔒 Information Security Policy 🤖 AI Policy 🧪 Secure Development Policy 🎯 Threat Modeling Policy ⚠️ Vulnerability Management 🏷️ Classification Framework

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

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