System Module
Documentation
Technical reference, integration landscape, and platform terms for the AnalyticalG simulation engine. This is an experimental research platform — read the disclaimer section before use.
On This Page
Overview
How It Works
Submit an Idea
Enter any text — a product claim, a tweet, a campaign message, a policy proposal. The engine preprocesses it for topic, valence, controversy level, and intent.
Synthetic Population Reacts
A population of 20–300 agents is generated from 12 archetypal personas, optionally conditioned on a country profile. Each agent carries 7 behavioral vectors: beliefs, identity, emotional triggers, social behavior, epistemological profile, and meta-narrative alignment. The most influential ~20% are powered by a live LLM tier; the rest run deterministic emergence mechanics.
Emergence Unfolds
Over multiple rounds, agents interact, influence each other, and shift stances. Thirty-one behavioral mechanics govern how beliefs move, spanning four research layers: behavioral economics (anchoring, identity entrenchment, hot-cold states), social influence (norms, fluency, authority, reciprocity), structural context (platform speed, institutional trust), and network science (complex contagion on a clustered graph). Agents hear their own neighbourhood, not the whole population. Where real-world grounding is available, the topic read is seeded from live signals rather than priors alone.
Results Surface
The engine outputs a decision-first verdict plus virality, polarization, and trust scores, opinion clusters, perception breakdown, and a full agent event feed — all available via the UI or API. Ensemble mode runs five seeded simulations and reports the spread as a confidence band (within-engine variance, not external validation).
API Reference
Endpoints
The simulation engine exposes a REST API. All simulations are asynchronous — you receive a job_id immediately and poll for results.
/simulate/intelligenceStart an /intelligence simulation. Returns job_id immediately. country conditions the population; seed makes the population reproducible.
{
"input": "Your idea or claim (3–500 chars)",
"agent_count": 100,
"rounds": 4,
"country": "US",
"seed": 42
}/simulate/ensembleRun 5 seeded /intelligence simulations of the same input and return all runs for a confidence band. The band reflects within-engine variance.
{
"input": "Your idea or claim (3–500 chars)",
"agent_count": 100,
"rounds": 4,
"base_seed": 42
}/simulate/architectureStart an /architecture simulation. Returns job_id immediately.
{
"input": "Your idea or campaign",
"agent_count": 100,
"ticks": 12
}/grounding?topic=...&claim=...What is actually circulating about a topic right now: stance distribution, themes, entities, tone, market signals. Returns pending while a cold build runs — poll again. Every response includes is_grounded; an ungrounded result is a first-class answer, never hidden.
/countriesList available country profiles for population conditioning.
/probe/counter-narrativesPROBE: generate 5 counter-narrative strategies and rank by neutralization effectiveness.
{
"input": "Narrative to stress-test",
"agent_count": 80,
"rounds": 3,
"country": "US"
}/decode/segmentsDECODE: run the same narrative against 1–3 persona-blend segments. Returns differential analysis.
{
"input": "Narrative to test",
"segments": [
{ "name": "Segment A", "persona_weights": { "tech_optimist": 2, "skeptic": 1 } }
],
"rounds": 3
}/decode/personasList all valid persona keys for DECODE segment builder.
/simulate/{job_id}Poll job status. Returns pending → running → complete (with result) or failed.
/healthLiveness check. Returns active job count and service status.
Rate Limits
10 simulation requests per minute per IP address. Exceeding this returns HTTP 429. The polling endpoint (GET /simulate/{job_id}) is not rate limited.
Integration Landscape
Potential API Integrations
AnalyticalG is designed to be a simulation layer that sits upstream of any content, campaign, or product decision. Integrations run in two directions: inbound grounding sources that feed real-world signal into the engine before a run, and outbound integrations that route simulation results into the tools where decisions happen.
Inbound — Grounding Sources Built Into the Engine
These are not roadmap items — they are connectors inside the Stage 4 ingestion pipeline. Each source flows through the same path: fetch → deduplicate → stance classification → topic clustering → a cached per-topic grounding that seeds the simulation's topic read. No personal data is stored; author identifiers are salted hashes and everything surfaces aggregate-only.
Global News Monitoring
GDELT
Worldwide news coverage, fetched keyless and deduplicated from syndication noise down to distinct stories. Powers the what's-circulating panel and the grounded topic read in production.
Use Case
Know what headlines, entities, and tone already surround your topic before you simulate against it.
Prediction Markets
Polymarket / Kalshi
Where a matched market exists, the market-implied probability enters the grounding as a signal of what people pricing real money believe. Price histories double as an external calibration backtest for engine trajectories.
Use Case
Anchor a simulation of a contested outcome to the market's live probability — and see how the engine tracks it.
Social Discourse (OAuth)
Posts plus comment reply graphs — real conversational structure, not just text. Connector complete; activation is credential-gated.
Use Case
Add opinionated discussion where people take positions, complementing news that merely reports them.
Real-Time Social (AT Protocol)
Bluesky
Real-time public discourse via authenticated AT Protocol access. Connector complete; activation is credential-gated.
Use Case
Capture the fast-moving, first-person layer of a developing narrative.
Video Discourse
YouTube Data API
Video titles, descriptions, and comment threads within daily quota budgets. Connector complete; activation is credential-gated.
Use Case
Reach the comment cultures where long-form narratives are argued out.
Population Realism
Survey Microdata (WVS / ESS / GSS / ANES)
Resample synthetic populations from real survey respondents instead of archetype blueprints, raked to target demographics. The next major realism upgrade.
Use Case
Populations whose trait distributions match a real society rather than an invented one.
How to Build an Integration
Every outbound integration below reduces to the same four-step pattern against the REST API — no SDK required, any HTTP client works:
- 01Submit — POST /simulate/intelligence (or /ensemble, /architecture, /probe, /decode) with your text. You get a job_id back immediately.
- 02Poll — GET /simulate/{job_id} every ~2 seconds until status is complete. Simulations typically finish in under two minutes.
- 03Consume — read the result JSON: verdict, scores, clusters, event feed. Map the fields your tool needs (e.g. polarization above a threshold blocks a publish queue).
- 04Optionally ground — GET /grounding for the same topic to attach the real-world context panel alongside the simulation result.
A Zapier zap, a Slack workflow, a GitHub Action, or a HubSpot workflow extension are all thin wrappers around this loop. Respect the rate limit (simulation starts are limited per IP; polling is not), treat scores as directional signals per the terms below, and surface the is_grounded flag wherever you display results.
Outbound — Where Simulation Results Could Flow
Marketing & Content Platforms
CRM / Marketing Automation
HubSpot
Push simulation scores (virality, trust, polarization) directly into HubSpot as campaign-level properties before launch.
Use Case
Pre-flight scoring of email campaigns, ad copy, and landing page messages before they reach real contacts.
Social Media Scheduling
Hootsuite / Buffer
Simulate a post's reception before scheduling. Block high-polarization content. Prioritise high-virality posts.
Use Case
Automated content moderation layer between draft and publish queue.
Knowledge Management
Notion / Confluence
Embed simulation results as structured blocks inside strategy documents and campaign briefs.
Use Case
Teams attach simulation evidence directly to decisions — auditable, contextual record.
Creative Tools
Canva / Adobe Express
Surface simulation scores alongside creative variants during the design phase.
Use Case
Designers see predicted audience response before the creative goes to review.
Analytics & Research
Web Analytics
Google Analytics 4
Compare simulation-predicted virality scores against actual post-launch engagement metrics. Close the feedback loop.
Use Case
Model calibration — use real-world data to improve simulation accuracy over time.
Social Listening
Brandwatch / Sprout Social
The engine already grounds topic reads from its own ingestion pipeline (see Inbound above). Commercial listening platforms could add brand-specific sentiment as a supplementary signal.
Use Case
Layer proprietary brand-tracking data on top of the built-in public-signal grounding.
Product Analytics
Mixpanel / Amplitude
Send simulation job metadata as events — correlate predicted scores with downstream product adoption.
Use Case
Product teams test feature messaging in simulation before rollout decisions.
Survey Platforms
Typeform / SurveyMonkey
Use simulation outputs to generate targeted survey hypotheses. Validate synthetic results against small human panels.
Use Case
Hybrid research workflow: simulate at scale, validate with humans at sample size.
Developer & Automation
Automation Platforms
Zapier / Make
Trigger simulations from any app event. Route results to Slack, email, Google Sheets, or any webhook.
Use Case
Non-technical teams run simulations from their existing tools without touching the API directly.
Team Communication
Slack
Post simulation results as formatted summaries to team channels when jobs complete.
Use Case
Strategy teams receive simulation reports in their existing workflow without context-switching.
CI/CD
GitHub Actions
Run simulations as part of content deployment pipelines — gate releases on simulation score thresholds.
Use Case
Automated content quality gate: high polarization score fails the pipeline.
LLM Orchestration
OpenAI / Anthropic API
Swap or augment the simulation's LLM tier with alternative models for specific persona types or domains.
Use Case
Domain-specific simulations using fine-tuned or specialised models alongside the default LLM tier.
Ecosystem
Collaboration Scope
AnalyticalG is positioned as a pre-deployment intelligence layer. These are the collaboration models under consideration.
Agency Partnerships
Marketing and communications agencies can white-label or API-integrate AnalyticalG into their client strategy workflows. Simulation outputs become billable deliverables — evidence-backed pre-launch risk assessments.
Research Institutions
Academic groups studying computational social science, opinion dynamics, or information diffusion can use the simulation engine as a research instrument. Custom persona configurations and raw output access available for institutional collaborators.
Product Teams
B2B SaaS products building content intelligence, brand safety, or audience understanding features can embed AnalyticalG as a simulation backend via API. OEM licensing under discussion.
Media & Publishing
Editorial teams, newsrooms, and content studios can use simulations to pressure-test stories, headlines, and framing choices before publication — understanding likely audience fractures in advance.
Policy & Public Affairs
Organisations working on policy communication, public consultation, or behavior change campaigns can simulate message reception across diverse synthetic populations before real-world deployment.
Interested in collaborating?
AnalyticalG is an experimental project by Truble Marketeer. Collaboration enquiries, integration partnerships, and research access requests can be directed to the team. This platform is in active development — early collaborators shape the product roadmap.
Legal
Terms of Use
1. Permitted Use
AnalyticalG is made available for research, strategic planning, content development, and educational purposes. You may use simulation outputs to inform decisions, generate insights, or support internal analysis. You may not use the platform to generate, validate, or amplify content intended to deceive, manipulate, or harm individuals or groups.
2. Simulation Output Nature
All outputs produced by AnalyticalG — including but not limited to virality scores, polarization indices, trust scores, agent reactions, opinion clusters, and adoption curves — are generated by a synthetic multi-agent simulation. They represent modelled probabilistic outcomes, not empirical measurements of real human behaviour. Outputs should be treated as directional signals, not definitive predictions.
3. Prohibited Use
- ✕Using simulation outputs to justify, validate, or execute disinformation campaigns.
- ✕Submitting inputs containing personally identifiable information (PII) of real individuals.
- ✕Automated bulk submission beyond published rate limits without prior written agreement.
- ✕Representing simulation outputs as primary research or empirical evidence in academic or legal contexts without explicit qualification.
- ✕Reverse-engineering, reselling, or redistributing the simulation engine or its outputs without authorisation.
4. Data & Privacy
Input text submitted through the platform is processed solely for the purpose of running the requested simulation. Inputs are not stored beyond the active job lifecycle (approximately 2 hours). No user accounts, personal data, or browsing histories are collected by the simulation engine. Standard server access logs may be maintained for operational purposes.
5. Intellectual Property
The AnalyticalG simulation engine, including its agent architecture, emergence mechanics, affinity matrix, and prompt system, is proprietary to Truble Marketeer. Simulation outputs generated from your inputs may be used by you freely for internal and commercial purposes, subject to the restrictions in Section 3.
Important
Experimental Project — Disclaimer
Experimental Research Platform
AnalyticalG is an experimental project. It is under active development and has not been validated against real-world behavioural data at scale. The simulation engine, its outputs, scores, and recommendations are provided strictly as exploratory research tools.
This platform is built and operated by Truble Marketeer. It is not a certified research instrument, a licensed decision-support system, or a regulated analytics product.
No Liability for Output Usage
AnalyticalG and Truble Marketeer accept no legal liability, responsibility, or accountability for any decision, action, harm, loss, or consequence arising from the use of simulation outputs. Users assume full and sole responsibility for how they interpret, apply, or act upon any result generated by this platform.
No Guarantee of Accuracy
Simulation outputs are probabilistic and stochastic by design. The same input may produce different scores across runs. Scores do not constitute predictions, guarantees, or warranties of real-world outcomes. AnalyticalG and Truble Marketeer make no representations regarding the accuracy, reliability, completeness, or fitness for any particular purpose of any simulation output.
No Professional Advice
Nothing on this platform constitutes legal, financial, medical, political, psychological, or professional advice of any kind. Simulation outputs are not a substitute for qualified expert judgement. Do not rely on them as the sole basis for consequential decisions.
Platform Availability
As an experimental platform, AnalyticalG may experience downtime, data loss, breaking changes, or complete discontinuation without prior notice. No SLA or uptime guarantee is offered. AnalyticalG and Truble Marketeer are not liable for any loss arising from platform unavailability or changes.
By Using This Platform You Acknowledge
- →AnalyticalG is an experimental, research-stage product.
- →All outputs are synthetic and probabilistic — not empirical measurements.
- →No liability for output usage rests on AnalyticalG or Truble Marketeer.
- →You assume full responsibility for any decision made using simulation results.
- →The platform may change, break, or be discontinued at any time.