Today at Pulsar we are pleased to unveil a new paradigm in identifying and neutralising brand crises before they escalate - Crisis Oracle. In this current hyper-connected digital landscape, a brand or organisation’s reputation can be challenged in an instant. Pulsar’s Crisis Oracle represents a step change in how organisations understand reputation risk: from reactive monitoring to predictive intelligence.
Designed to detect, interpret, and escalate emerging brand risk before it becomes a crisis, this groundbreaking agentic AI system has been engineered for predictive brand crisis intelligence. This sophisticated solution moves beyond simple keyword tracking to provide a true understanding of narrative momentum. By identifying the velocity, visibility, and contagion of negative conversations, it empowers brands to predict and mitigate crises with unprecedented accuracy.
What Makes Crisis Oracle Different?
Traditional social monitoring tools, which rely on basic keyword tracking and sentiment analysis, are no longer sufficient to navigate the complexities of the modern information environment. They often generate significant noise, miss contextual nuance, and fail to provide a forward-looking view of emerging crises.
Built on a foundation of Agentic AI, Crisis Oracle moves beyond simple monitoring to offer a multi-dimensional view of brand health. It is engineered to understand the momentum of negative narratives; how they form, accelerate, and spread across networks. By detecting these critical shifts early, the system empowers teams to predict, anticipate and respond strategically and effectively, neutralizing threats before they escalate into full-blown crises.
Here’s how it stands apart:
Narrative-First Approach: Instead of tracking individual posts, the Crisis Oracle clusters conversations to identify emerging narratives, providing a holistic view of brand-related discussions.
Momentum Scoring (θ): Our proprietary momentum score (θ) combines the size, visibility, and velocity of conversations to provide a quantifiable measure of narrative risk.
Expert Risk Assessment: The Referee, an AI reasoner, reviews each narrative and provides human-readable concise summaries of each narrative, making it easy to understand the context and severity of the situation.
Actionable Alerts: Receive alerts only when a narrative’s status changes, along with a daily digest that summarises key developments, eliminating alert fatigue.
Multilingual and Multi-Source: Brand Crisis Oracle monitors conversations across X, News, and Blogs, etc. ensuring a comprehensive view of the digital landscape.
How The Crisis Oracle Works
Crisis Oracle is powered by P.U.L.S.E™ (Persistent Upshift in Latent Signal Emergence), Pulsar’s proprietary momentum scoring agent for brand risk. Rather than asking “Is sentiment negative?” P.U.L.S.E asks a more important question: Is this narrative gaining dangerous momentum?
P.U.L.S.E calculates a momentum score (θ) for each narrative by combining three signals:
Size: how much negative conversation exists
Visibility: how exposed audiences are to it
Velocity: how fast it is accelerating
This produces a stable, comparable measure of narrative risk; one that reflects trajectory, not just volume. P.U.L.S.E then maps that momentum onto clear escalation states: Calm → Concern → Incident → Crisis, with built-in persistence to avoid alert fatigue.
Below is an outline of the sophisticated, multi-step architecture employed by the Crisis Oracle to deliver unparalleled insights:
Ingestion: Hourly ingestion of brand-related conversations from sources like X, News sites, and Blogs, regardless of language (can handle 150+ languages).
Context Hydration: To ensure full understanding, the system fetches the text of any parent or quoted posts referenced in the initial data, creating a lightweight context map.
Brand-Target Stance Gate: This is the first critical filter. A proprietary NLP-based classifier filters for posts that are truly about the brand and assesses the stance (hostile, critical, neutral, or positive) and intensity of the conversation.
NEG Gate & Bundling: Posts that pass a negative threshold are bundled with their related context posts (replies, shares, etc.).
Weighted Joint Clustering: Negative posts and their context are clustered together, with negative posts acting as the “anchor” to ensure that the crisis signal is not diluted by neutral or positive context.
24-Hour Rolling Store: A sliding 24-hour memory of all posts and their connections is maintained for continuity, allowing the system to stitch hourly data into a coherent narrative.
P.U.L.S.E Scoring: For each narrative cluster, P.U.L.S.E calculates the volume, visibility, and velocity of negative, brand-targeted posts.
Escalation Ladder: The momentum score (θ) is used to assign a stable state to each narrative; Calm, Concern, Incident, or Crisis, with built-in hysteresis logic to prevent alert fatigue.
Referee Agent: An LLM-based agent provides a final, human-like assessment of the situation, nudging the stage up or down as needed and providing a concise summary.
Alerting & Digest: The system sends hourly alerts via Slack or email only when a narrative’s state changes and provides a daily digest that rolls up all active narratives.
The Outcome
With Pulsar’s Crisis Oracle, brands and their teams can:
Predict emerging brand threats before they become full-blown crises.
Gain a deep understanding of the narratives shaping their brand’s reputation.
Reduce false positives and focus on the threats that truly matter.
Benefit from human-level judgment at machine scale, enabling teams to move from reactive to predictive reputation intelligence.
Built on TeamMates: Pulsar’s Agentic AI Platform
Whilst Crisis oracle will be the second TeamMate that we are launching this year, it is the first Oracle-class Insight Agent within Pulsar TeamMates, our agentic AI layer designed to transform Pulsar from a data platform into an operating system for your audience. TeamMates are specialised AI agents trained to perform complete insight, PR, and marketing jobs autonomously, working continuously across Pulsar’s live datasets and reasoning layer. They don’t just surface data; they apply judgment.
Within TeamMates:
Sentinels detect and protect in real time
Oracles predict and anticipate
Analysts research and explain
Custodians govern and enforce
Crisis Oracle is the first agentic system of the Oracle class: agents designed for prediction, escalation, and scenario awareness, bringing foresight directly into PR and communications workflows.
Availability & Access
If you’re a TRAC user and interested in getting all setup with Crisis Oracle, then reach out to your Account Manager, or simply message us via the in-app chat. Setup is easy and straightforward, with options to specify your preferred alert channels. Currently we support email alerts or Slack channel delivery, but other channels can be supported too.