Expand Investigative Horizons with Agentic AI
Criminals and terrorists no longer work at human speed. Nation-state adversaries, organized criminal networks, hostile foreign intelligence services and workaday scam artists now deploy artificial intelligence to conduct operations at scale. They use AI to:
- Run disinformation and influence campaigns
- Manufacture synthetic identities,
- Probe supply chain vulnerabilities
- Automate reconnaissance
- Write and deploy malware for accessing computer systems
- Search scientific literature for information needed to accelerate biological- and chemical-weapons campaigns
- Map critical components of national infrastructures, identifying weaknesses
- Commit other malfeasance
To meet the challenges of an AI-powered crimescape, those working in national and corporate defense and intelligence need to fight fire with fire. They must deploy new technologies for AI-powered decision making. In doing so, they can improve situational awareness, threat intelligence, identity risk intelligence, and vendor risk intelligence — obtaining insight quickly, and at scale. In the best of these systems, AI collaborates with human practitioners, helping them obtain news depths of insight.
Learn the basics
Data Sheet
Babel Street Insights Investigator
Data Sheet
Babel Street Insights
Product brief highlighting benefits and features of Babel Street Insights for People Search, Document Search, and maximizing the use of publicly available information.
Infographic
Scaling Expert Tradecraft with Agentic AI
Case Study
Babel Street Insights Investigator: Empowering Rapid, Secure Intelligence Operations at CIRAT
Overcoming human limitations
Today, intelligence is increasingly based on analyses of publicly available information (PAI). The amount of PAI available for study is almost unimaginably vast. Social media posts and shipping information. Commercially available datasets and dark-web marketplace sales. Legitimate news media and blogs. Government records, vehicle telemetry, and cell site-location information. The world creates hundreds of terabytes of new information every day.
No matter how skilled, human analysts can’t keep up on their own. They face several analytical challenges.
The first of these is data gathering. Governments and enterprises spend too much time on inefficient, manual processes for gathering and collating the type of data needed for investigative insight.
The second is a dependence on target-first or hypothesis-driven searches. These types of searches can only return insight on situations an analyst either knows to be, or suspects of being, problematic. If he cannot imagine a situation, he cannot glean insight into it.
What does this mean?
Consider the following scenario. You’re planning an expensive family outing to a large theme park. You’ve been there a few times as an adult, so you think you know the drill. Based on your existing knowledge, you decide to check the park’s height requirements. (Your son is short, and you don’t want to go if he’ll have to sit out most of the rides.) You check the weather, the traffic. You search historical attendance records: Who wants to spend the whole day waiting on line?
Everything looks good, so you go. You get to the park, and find it closed.
Why? Since your last visit to the theme park, a zoo has opened nearby. Officials have three times cited the facility for failure to properly contain its animals. Today, an orangutan has escaped, entered the theme park, and is climbing all the rides. It will take hours to subdue and capture the animal.
You didn’t think to search for this. No one would think to search for this. But a simple query to an AI decision making system — Tell me about any issues in the theme park and surrounding environment that might negatively impact a visit by two adults and a seven-year-old — would have uncovered this threat.
The third challenge faced by investigators is finding the time to apply their professional expertise and crucial insight to information uncovered. The amount of data they’re required to review, coupled with the number of reports they’re expected to compile, make it difficult to devote time to this fundamental investigative step.
The AI solution
To overcome human limitations, analysts and others need machine-scale solutions offering collaborative, agentic AI built on expert investigative tradecraft. AI agents should work in tandem to tackle data collection, analysis, and reporting tasks — seamlessly moving from search functions to work functions. To satisfy the requirements of high-stakes investigations, AI solutions must be built on world class data.
Systems built on governed, collaborative AI agents can help analysts:
Collaborative AI agents can help broaden investigative paths for a broader view of possible threats. In essence asking an analyst “Have you considered…?” they help expand certain areas of investigation while quickly shutting down others as erroneous.
Consider this. An IRS investigator is researching a possible instance of tax evasion. The target company has shell subsidiaries in multiple countries. It also has a history of unusually routed international wire transfers, and financial intermediaries (banks, brokers) working in jurisdictions known for weak financial oversight. The investigator may believe he’s found evidence of tax evasion.
However, collaborative AI agents have detected additional information. Executive travel to conflict regions. Donations to charities suspected of laundering money. Connections or connections-of-connections to terrorism-related social networks. An AI system built on collaborative agents may prod the investigator to consider whether this company isn’t just evading taxes, but financing terrorism.
In a real-life example, experienced analysts conducted open-source research into solar panel companies. Looking for potential links to adversarial nations, they followed standard, hypothesis-driven tradecraft. This approach overlooked broader vulnerabilities within the solar panel industry. Working with Babel Street, analysts asked their collaborative AI solution to holistically analyze an entire body of collected research. Babel Street surfaced financial- and industry-level risks that the team had not previously considered and suggested new queries for reshaping the investigation.
Conversely, collaborative AI agents may quickly flag certain investigations as erroneous. For example, a government analyst searching for domestic terrorists may find an online community using maps, countdowns, obscure symbols, coded language, and talk of “operations.” The analyst may feel he’s found a terrorist cell. Collaborative AI may point out the possibility that this group is discussing a multiplayer roleplaying game.
The data foundation
No AI-powered insight means anything if the intelligence uncovered is based on bad data. Out-of-date data. Disconnected data. Data with blind spots.
The best AI solutions are trained on the best available data. They can search data originating from countries around the globe, created in multiple languages. They refine, standardize, and enhance this data for operational use. These solutions also access and store ephemeral data: social media posts, discussion-forum comments, and other communications that may be deleted by the creator or the platform. This type of data can provide valuable insight when properly stored and processed.
Benefits
Working with modern collaborative AI systems, government, military, and intelligence agencies — along with large enterprises — often obtain:
- Faster, more accurate understanding of organizational risk and exposure
- Earlier identification of risk indicators
- Quicker orientation during emerging or rapidly evolving situations
- Time to prepare for, respond to, and mitigate threats
- Faster and more complete identity and situational assessments
- Earlier detection of hidden relationships
- Earlier detection of compliance violations
- Stronger confidence in vendor, partner, and investment decisions
- Defensible outcomes that stand up to regulatory, legal, and executive scrutiny
Why Babel Street?
Babel Street, a global leader in mission-grade risk intelligence, has spent more than a decade providing military, defense, intelligence, and enterprise communities with AI-powered insight based on world-class data. That experience and expertise have shaped the Babel Street Agentic Risk Intelligence Platform. Setting a new standard for threat detection, the platform helps governments and businesses outpace dynamic threats by providing panoramic, up-to-the-second insight into geopolitical risk, insider threats, supply chain vulnerabilities, and more.
Insights Investigator is an agentic capability that puts tradecraft-trained AI agents at the forefront of investigative work. Built for high-stakes environments, Insights Investigator empowers a shift from target-first or hypothesis-driven searches to more holistic investigations. These investigations are directed by human practitioners and executed by multiple AI agents working in tandem.
Insights Investigator is differentiated by:
- Elite tradecraft: Investigator is modeled on proven investigative workflows and techniques used by expert analysts and investigators. Human practitioners remain in control of investigative objectives, pathways, scopes, and outcomes.
- Agentic execution: Rather than merely returning chatbot-like answers to queries, Insights Investigator’s AI agents execute multi-step investigative tasks.
- Data Dominance™ — Babel Street’s rights-cleared, mission-curated, multilingual data pipeline underpins Investigator. We cull this data from information published in more than 200 languages. Our entity extraction capabilities structure data at ingestion, improving search precision and enabling more consistent, evidence-backed analyses.
- Trust, Governance, and Auditability: Unlike black-box AI systems, Investigator offers explainable AI: delineating its own research plans, query logic, reasoning, and sourcing. Investigator also provides audit trails and other types of traceability that satisfy regulators, leadership, and stakeholders.
At Babel Street, we fuse elite tradecraft with the world’s deepest collection of rights-cleared data to empower insightful, defensible decision-making in the AI-on-AI era. In an uncertain economic and geopolitical climate, the Babel Street Agentic Risk Intelligence Platform helps military, defense, intelligence, and enterprise organizations make the type of fast, fully informed decisions needed to protect themselves and the communities they serve.
Frequently Asked Questions
What is AI decision making and how does it work?
AI decision making is the process by which an AI system analyzes information, weighs options, and recommends actions. An AI collaborator receives input (typically gleaned from publicly available information), processes it, and evaluates it before presenting its findings to a human practitioner.
What is the difference between human decision making and AI decision making?
Human decision making is based on experience, judgment, ethics, context, social nuances, and other factors. AI decision making is based on data, mathematical models, and probabilities. It depends on the system’s ability to spot and understand patterns and statistical relationships. Bad data, bad model design, and training bias can all harm AI’s ability to make good decisions.
How is AI decision making used in modern governments and businesses?
AI-driven decision making is used in modern governments and businesses to analyze data, identify patterns, predict risk, predict outcomes, and help professionals make faster, better-informed decisions. It is a valuable tool for situational risk analysis, fraud detection, law enforcement investigations, supply chain monitoring, cybersecurity, financial forecasting, compliance monitoring, and other tasks.
How can governments and companies integrate AI decision making into their existing workflows?
Rather than trying to replace entire processes at once, many governments and companies gradually integrate AI decision making into their workflows — initially as a support layer, then as agentic systems that can perform complex tasks with minimal human involvement. Many governments and businesses start by deploying AI decision making for tasks such as threat triage, fraud risk scoring, and compliance scoring.
What technologies enable AI decision making systems?
A variety of different technologies work in tandem to enable AI decision making systems. These include machine learning, deep learning, natural language processing, entity resolution, and predictive analytics programs, along with neural networks and large language models.
What are the best practices for implementing AI decision making solutions?
Best practices for implementing AI decision-making systems usually focus on balancing automation, accuracy, governance, and human oversight. When implementing AI for decision making, organizations should clearly define the problems they want AI to help solve; choose AI systems built on high-quality, well-governed data; test before deployment; and monitor continuously after deployment.
What are the risks associated with AI decision making systems?
The risks associated with AI decision making systems include AI incorrectly flagging innocent activity as suspicious, while missing signs indicative of legitimate dangers. If the datasets AI systems are trained on are in any way biased, the AI system can inherit that bias. This can lead to unfair outcomes in hiring, policing, lending, and other practices. AI systems can also “hallucinate”: inventing fake facts, misstating evidence, and creating nonexistent relationships.
What are the challenges of implementing AI decision making in government agencies and businesses?
The challenges of implementing AI decision making in government and business include suboptimal data quality that can lead to incomplete or incorrect insight, bias, and a lack of transparency. Organizations may face ethical and reputational risks if AI makes discriminatory or unsafe recommendations or disseminates misinformation. Organizations can face problems integrating AI systems with older software already in use. Implementation costs can be high.
What is the role of machine learning in AI decision making?
Machine learning is at the heart of AI decision making. It is the system component that, over time, extrapolates patterns from data. Based on thousands or even millions of examples, machine learning systems can learn that Pattern A tends to correlate with Outcome B. This capability empowers AI systems to make probabilistic judgments in complex situations.
How does AI decision making help organizations analyze large datasets?
AI decision making helps organizations analyze large datasets by processing information faster, more consistently, and at a much larger scale than human analysts alone could manage. These systems also provide analyses of unstructured data, data integration, anomaly detection, and predictive insights, among other capabilities.
What are real-world examples of AI decision making in action?
Currently, AI decision making systems are used by a broad variety of organizations for a plethora of tasks. Law enforcement uses AI decision making systems to identify criminal networks and prioritize investigative leads. Military and intelligence agencies use them for situational awareness and identity intelligence. Corporate security organizations use them to detect insider threats, and the type of unusual network behavior that can indicate unauthorized access.
How does AI decision making improve threat intelligence analysis?
AI decision making improves threat intelligence analysis by finding, connecting, and correlating information across massive datasets. It processes more information faster, and with greater pattern recognition, than manual analysis alone can.
What role does AI decision making play in risk management?
AI decision making plays a major role in modern risk management. It helps organizations identify, assess, prioritize, and respond to potential threats quickly, and at scale. It accomplishes this through real-time threat monitoring, detection of hidden risk patterns, predictions of future risk, and threat prioritization.
How can AI decision making support cybersecurity strategies?
AI decision making supports cybersecurity strategies by helping organizations analyze massive amounts of security data that human teams cannot process on their own. In doing so, AI decision making systems detect threats faster, and help automate defensive actions.
Disclaimer:
All names, companies, and incidents portrayed in this document are fictitious. No identification with actual persons (living or deceased), places, companies, and products are intended or should be inferred.