Veris Truth Engine

Veris Ecosystem is a full-stack platform for an autonomous investigative AI agent, using a Python/LangGraph backend and Next.js frontend to enable AI-driven investigation, reasoning, and interactive visualization of evidence and insights.

Watch

Tech Stack

Uncovering the Truth with Autonomous AI

In an era of deepfakes, misinformation, and overwhelming data sprawl, manual investigation is becoming impossibly slow. Veris was built to flip the script.

Developed for Recurzive V2, the Veris Ecosystem is not just a search engine—it is an autonomous investigative AI agent. It doesn't just fetch links; it actively reads, verifies, builds knowledge graphs, and iteratively reasons through complex queries until it finds the truth.

The Cognitive Engine: LangGraph Backend

The heart of Veris is its backend intelligence, built with Python and LangGraph. Instead of a simple prompt-response loop, we engineered a stateful, looped pipeline controlled by an AgentState object. This allows the AI to "think" in steps, mimicking a human detective:

1. The Collector: The agent begins by scouring online sources, actively fetching raw data and potential evidence related to the user's query.
2. The Verifier: Not all data is equal. This node acts as a critical filter, cross-referencing information, evaluating credibility, and assigning trust scores to the evidence.
3. The Graph Node: As verified facts emerge, the system synthesizes them into a relationship map, connecting entities, events, and evidence logically.
4. The Refiner: The AI evaluates its own findings. Do we have enough to conclude the investigation? If not, it formulates a next_query and loops back to the Collector. If yes, it finalizes the investigation.
5. The Reporter: Finally, the agent compiles the structured data into a comprehensive, readable report.

This autonomous looping means the agent can go down rabbit holes, hit dead ends, pivot, and ultimately return with a highly verified conclusion.

Gallery

Team Contributions

Nikhil R

Full Stack Engineer

Bridged the LangGraph-powered backend with the Next.js frontend, orchestrating real-time state synchronization and designing the agent's iterative reasoning and self-evaluation workflow.

Next.jsPythonLangGraph

Aryan Mishra

UI/UX & Frontend

Designed the command-center aesthetic, interactive graphs, and the particle animations for the dashboard.

Next.jsTailwind CSS

Prajwal S

AI Engineer

Implemented and collaborated on the Verifier and Refiner nodes in the LangGraph pipeline, enabling evidence verification and iterative investigation through self-evaluation and query refinement.

PythonLangGraph

Nishanth Antony

AI Architect & AI Engineer

Designed and implemented the AI architecture for the Veris investigative agent, including the reasoning workflow, agent state design, and integration of verification and iterative investigation loops.

PythonLangGraphLLM Orchestration