Introducing Adarga’s RFI Assistant: Harnessing the Power of Agentic AI

In the fast-paced world of intelligence analysis, speed and accuracy are everything. Adarga is at the forefront of empowering intelligence analysts to maximise their productivity by leveraging cutting-edge GenAI and NLP technologies. This relentless drive to push the boundaries of the possible led to the creation of Adarga’s RFI Assistant - a rapid innovation project designed to transform the way intelligence analysts respond to RFIs (Requests For Information).

The project was born from a conversation between an Adarga data scientist and a former military intelligence analyst. They were discussing a persistent challenge: the time-consuming nature of answering RFIs. Responding to RFIs involves gathering comprehensive intelligence on topics such as the operational activities, leadership, and logistics of groups like the proscribed organisation “Al Shabab.”

Traditionally, RFIs require analysts to sift through vast amounts of OSINT and in-house data to produce a response. The question was simple: could this be done better?

   

What is Adarga's RFI Assistant?

Forged in Vanguard – Adarga’s AI Capability Accelerator – the RFI Assistant is designed to drastically reduce the time it takes to answer an RFI, taking advantage of AI while keeping humans firmly in control. Powered by Adarga’s proprietary AI technologies, the RFI Assistant uses a sophisticated agentic workflow built on top of the Adarga Knowledge Platform (AKP), using our own carefully crafted custom agents.

The concept of an agentic workflow is a radical departure from traditional AI processes. Unlike conventional models that generate outputs in a single pass, agentic workflows involve iterative, interactive AI development where AI agents engage in self-reflective processes, all while enabling human oversight and control at every step.

Agentic AI allows for systems of self-managing and correcting AIs to interact with one another to amplify their strengths while mitigating their weaknesses. These autonomously interacting AI systems are dramatically more powerful and capable than previous technologies.

   

How are Adarga’s custom AI agents transforming RFI responses?

The power of Adarga’s RFI Assistant lies in its bespoke AI agents; each optimised for a specific phase of the RFI response process. By leveraging Adarga’s specialist domain knowledge, deep technical expertise, and curated geopolitical data sets, we are uniquely able to craft expert agents that are optimised for the task of completing Intelligence Analyst RFIs.

The agents behind the Adarga RFI Assistant are equipped with a range of AI-driven capabilities from Adarga’s proprietary Knowledge Platform – our ecosystem of 35+ models geared for information intelligence use cases. Features such as document searches, entity extraction, event identification, and relationship mapping are unlocked for the agents, enabling the RFI responses to be compiled using highly relevant information.

   

Adarga's Agentic Workflow

1. RFI Receipt and Ingestion: The process begins when the RFI is received, typically in a text-based format like a Word document. The content is then ingested into the AKP, and relevant details are extracted to transform the RFI into an internal format for subsequent processing by Adarga’s RFI Assistant.

2. Planning: The Planner Agent takes the lead by analysing the RFI and outlining the steps needed to generate a response. This is an iterative process, with an intelligence analyst refining the steps to ensure accuracy.

3. Task Management: The Manager Agent assigns tasks to subsequent agents based on the planned steps. It decides how to allocate resources efficiently to complete the RFI.

4. Analysis and Research: The Analyst Agents are the workhorses, leveraging Adarga’s search and retrieval capabilities to drill down into the research, iteratively narrowing the scope to deliver precise findings. Analyst agents can either be created as broad generalists or as specialists. Generalist agents have access to a broad set of capabilities from the AKP. To improve the quality of responses in a specific domain, specialist analyst agents can be created with a focus on specific fields such as Chinese foreign policy, national security, or other areas. Crafting these bespoke agents requires deep domain knowledge. A specialist agent on the Middle East, crafted by Adarga, was enabled for the demonstration videos presented here.

5. Review and Validation: Finally, the Reviewer Agent cross-verifies the outputs, ensuring all citations are accurate and relevant to the RFI. This step is crucial for maintaining the integrity of the analysis. It acts as a guardrail to mitigate hallucinations and ensure AI safety. 

6. Human Oversight: Throughout the process, human analysts are engaged to review, validate,  and refine outputs at each stage, maintaining control over the final deliverable. By ensuring human control over AI agents, the Adarga RFI Assistant generates RFI responses that are of significantly higher quality than either humans or AI could produce on their own.

   

Driving unprecedented speed and efficiency

The result? Information at the speed of relevance, curated by specialist agents. Adarga’s RFI Assistant can generate outputs for intelligence analysts in mere minutes - a task that would traditionally take a week to complete. For less experienced analysts, the process could stretch even longer, with extensive validation required by subject matter experts.

Most importantly, the RFI Assistant keeps the human analyst firmly in control.

   

Rapid innovation, underpinned by the Adarga Knowledge Platform

Adarga’s RFI Assistant is just one demonstration of Adarga’s rapid innovation capabilities which leverage the robust AI and engineering toolsets of our proprietary Knowledge Platform. Teamed with our AI Capability Accelerator, Vanguard, we can quickly turn ideas into real-world impact.

By combining the efficiency of AI with human domain expertise, Adarga’s RFI Assistant represents the potential future of AI-supported intelligence analysis: faster, smarter, and more accurate.

   

To find out more about Adarga’s rapid innovation capabilities and applied AI services, get in touch here. 

   

About the Adarga Knowledge Platform

Adarga’s Knowledge Platform (AKP) is designed to ingest, process, and extract insights from vast amounts of open-source and in-house information in over 75 languages. It underpins all of Adarga's products and services, including our flagship product Vantage.

By integrating 35+ machine learning, AI, and NLP models – including classical models, bespoke solutions, and large language models with robust guardrails – the AKP addresses complex, domain-specific Information Intelligence use cases.

Teamed with our Vanguard technology, processes, and people, AKP allows for rapid innovation and implementation of advanced innovative technologies - ensuring that Adarga leads the way in this incredibly fast-moving domain.