How Monks uses Amazon Nova to make complex brand compliance decisions in real time

Organizations often struggle to turn complex brand or policy guidelines into clear, consistent answers — especially at scale. Often these documents have nuanced information, or even contradictory suggestions, making it harder for generative AI solutions to provide reliable answers. The challenge was to build a chatbot that could deliver trustworthy, real-time guidance across different tasks and user intents, without incurring the high latency and cost associated with other large language models.

Monks developed a brand guideline chatbot that can simultaneously digest up to 20 different documents to create high-fidelity guidance for global marketing teams. The chatbot uses a blend of Amazon Nova models to balance speed, accuracy and cost. Nova Lite is used to create structured research plans and judge document relevance. Nova Pro is used when deeper analysis is needed, leveraging XML prompts and chain-of-thought techniques. Each document is then processed by Nova Lite or Pro, and the most relevant insights are merged into a final response. By matching each task to the right model, Monks delivers fast, accurate results at a fraction of the cost of traditional LLM workflows.