The latest articles on Forem by adityaprasad-sudo (@adityaprasadsudo). https://forem.com/adityaprasadsudo https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3740200%2F4edba1d9-9ec4-406d-81ca-0e0baa8a6ab1.png https://forem.com/adityaprasadsudo en adityaprasad-sudo Thu, 29 Jan 2026 17:38:00 +0000 https://forem.com/adityaprasadsudo/building-a-triple-failover-rag-why-one-llm-wasnt-enough-for-singaporean-law-1kh6 https://forem.com/adityaprasadsudo/building-a-triple-failover-rag-why-one-llm-wasnt-enough-for-singaporean-law-1kh6 <p><strong>The Hook (The Problem)</strong> I’ve always been interested in the relationship between code and <strong>“The Code”</strong> – the legal systems that govern our world. In Singapore, the laws are specific, but for the layman or legal researcher, wading through thick legislation and past changes can be like <strong>looking for a needle in a digital haystack.</strong></p> <p>I chose to create a <strong>custom RAG (Retrieval-Augmented Generation)</strong> engine to address this problem. But I soon realized that when working with legal data, <strong>hallucinations are not just frustrating – they can be costly.</strong></p> <p><strong>The Technical Challenge (Solving RAG Hallucinations in LegalTech)</strong> Traditional <strong>RAG architectures are brittle</strong>. What if your main model goes down or your context window gets congested? The system breaks. I wanted to create something** that was “Apple-level” in its polish but “Industrial-grade” in its robustness.**</p> <p><strong>This prompted me to create a Triple-AI Failover Backend.</strong></p> <p><strong>Gemini</strong> I have this my primary AI in my webpage.</p> <p><strong>Llama 3</strong> If Gemini hits the rate limit the webpage AI seamlessly switches to Llama 3.</p> <p><strong>Groq</strong> If both of those two fail it switches to Groq.</p> <p>As the engine uses <strong>FAISS</strong> for semantic embeddings, it not only searches for keywords but also comprehends the meaning of <strong>Singaporean legislative language</strong>.</p> <p><strong>The “Aha!” Moment</strong> The moment of truth wasn’t only in the search results but also in the UI. I shifted away from the “chatbox” style and opted for a high-precision, <strong>interactive UI that resembles more of a research tool than a playground.</strong></p> <p>Check Out the Project I have decided to open-source the engine. Whether you are interested in <strong>LegalTech</strong>, <strong>analyzing RAG architectures</strong>, or simply want to see how a <strong>multi-model failover is implemented in Python, I would be delighted if you took a look.</strong></p> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fj50uoym5hkhx4lrtqz9f.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fj50uoym5hkhx4lrtqz9f.png" alt=" " width="800" height="556"></a>]</p> <p>Live Experience: [<a href="https://adityaprasad-sudo.github.io/Explore-Singapore/" rel="noopener noreferrer">https://adityaprasad-sudo.github.io/Explore-Singapore/</a>]</p> <p>Deep Dive into the Code: [<a href="https://github.com/adityaprasad-sudo/Explore-Singapore" rel="noopener noreferrer">https://github.com/adityaprasad-sudo/Explore-Singapore</a>]</p> <p>The Future This is only the beginning. As Singapore’s policies change, so will the engine. I am currently researching [How to build a accurate text embedding model] to further increase the autonomy of the research.</p> ai python opensource rag