Inspiration
The inspiration behind the project has to be the terrible consequences of misinformation, such as division, strife, and even scams. As information velocity increases and generative tools lower the barrier to content creation, distinguishing truth from falsehood is becoming harder for everyday users. Veritas AI is motivated by the need for a system that does not merely assert correctness but exposes evidence, sources, and reasoning so users can independently verify claims.
What it does
The app basically accepts claims and verifies those claims based on publicly available information. It also highlights the sources it cites to verify those claims. The essence is to provide verifiable information
How we built it
The application was prototyped using Google AI Studio, leveraging large language models to interpret user-submitted claims, retrieve relevant public information, and synthesize evidence-backed responses.
Challenges we ran into
The primary challenge was time. Starting the project late compressed experimentation, evaluation, and iteration into a narrow window. This constrained the breadth of retrieval strategies tested and limited deeper exploration of confidence scoring and source weighting.
Accomplishments that we're proud of
We are proud to build such an important app in a small amount of time
What we learned
What's next for Veritas AI
The next phase of Veritas AI focuses on operational maturity and epistemic rigor. On the engineering side, we plan to introduce CI/CD pipelines to continuously test retrieval accuracy, citation validity, and regression in reasoning quality as models or prompts evolve. Automated evaluation using curated truth benchmarks and adversarial misinformation datasets will ensure consistent performance over time.
Beyond infrastructure, we aim to explore new methods for approximating practical absolute truth in news contexts. This includes multi-source consensus modeling, temporal consistency checks (tracking how claims evolve over time), source credibility weighting, and confidence intervals instead of categorical judgments. The long-term vision is a system that treats truth as evidence-weighted, time-aware, and continuously updated. To build a ubiquitous news verification tool and unique methods that are useful for Nigerians, and the world at large
Log in or sign up for Devpost to join the conversation.