One platform across collection, context and action.
A single workflow from signal to reviewed action. Built on open-source components, with a simple rule: nothing leaves the system without a person approving it.
Narrative intelligence
Coordinated framing on EU policy position
Policy lead asking who covers EU narrative analysis
What Eidoscore does
Each capability owns a job. Together they turn a public signal into a reviewed action with evidence attached.
Collect
Prompt monitor
Run defined prompts across selected AI engines on schedule. Store responses, ranking, citations and change over time.
Telegram intelligence
Monitor channels, classify messages and detect misinformation and emerging topics.
Reddit intelligence
Subreddit-level signal: sentiment, hostile framing, support, questions, reviews and opportunities.
Contextualise
AI visibility dashboard
Where the brand appears, whether it's recommended, which competitors show up, and how visibility changes by prompt, model and market.
AI perception analysis
Tone, clarity, credibility, source quality, recency, confidence, risk and narrative accuracy.
Source and citation intelligence
Identify sources used by AI models. Surface missing, weak or outdated sources and the actions to improve them.
Cross-channel narratives
Group similar mentions across sources into narratives. Track whether a story is growing, fading or moving between platforms.
Misinformation and risk
Flag likely misinformation, hostile campaigns, manipulated framing and high-risk claims for human review.
Act and report
Action centre
Turn insights into tasks: create content, update a source page, respond, escalate, prepare a statement or monitor.
Human review and approval
Routes alerts and drafted responses to the review channel or dashboard for approval before any external action.
Publishing and reporting
Approved responses flow to publishing. Reports and evidence packs are stored and exported.
Four scores. One consistent way to prioritise.
Every insight carries the same four scores, so routing and decisions stay explainable across the team.
Harm potential, misinformation likelihood, public visibility and velocity.
Evidence strength, source reliability and model certainty.
Speed of spread, stakeholder sensitivity and time relevance.
Lead potential, content angle and supporter engagement.
Modular by design, so the durable value isn’t a single vendor.
Channel rules change. Models change. Eidoscore’s collectors are replaceable; the durable value sits in normalisation, classification, source traceability, the dashboard and the workflow.
Collection
One connector per channel. Telegram, Reddit, AI answer engines, web and news in phase one. Replaceable as platform rules evolve.
Contextual analysis
Normalise and store. A language model classifies each item for topic, stance, type, weight and early narrative signal.
Action and output
Dashboard, alerts, AI-drafted responses held for approval, optional handoff to publishing, lead signals to CRM.
Orchestration and governance
Public data only, recorded lawful basis, retention limits, EU hosting, single sign-on, secret storage and full audit trail.
Three proof points first.
Channel breadth comes later. The MVP proves the intelligence layer where the value is hardest to fake.
AI answer engines
Run a defined set of prompts across multiple models. Store responses, ranks, citations and visibility over time.
Telegram
Monitor approved channels. Classify messages. Detect repeated or risky claims and emerging topics.
Track relevant subreddits for sentiment, framing, support, questions, reviews and live opportunities.
More channels are added over time, only where platform terms allow and a compliant route exists.