BluntlyAI

#problem

BluntlyAI is an AI data analyst product that ingests complex datasets and produces clear, actionable insights and automated reports. It addresses the gap between raw data and decision-ready analysis by coordinating multiple large language models through OpenAI to interpret data and generate written summaries. The system pairs a NestJS API layer with a Next.js front end and runs on AWS.

#what-i-built

• ▸I architected the BluntlyAI platform end to end, designing the NestJS backend, the Next.js front end, and the AWS deployment.
• ▸I built the NestJS services that orchestrate multiple LLMs via OpenAI to analyze datasets and generate automated reports.
• ▸I designed the data analysis pipeline that converts uploaded datasets into structured insights and written report output.
• ▸I developed the Next.js interface where users submit data, review generated insights, and access automated reports.
• ▸I deployed and operated the application on AWS, owning the cloud infrastructure and release process.

#stack

#outcome

Production multi-LLM AI data analyst platform live at bluntly-ai.com — full-stack ownership across the NestJS analysis backend, OpenAI multi-model orchestration layer, and AWS deployment.

#key-decisions

• 1.Orchestrated multiple LLMs through OpenAI rather than a single model, routing summarization and pattern-detection subtasks to the best-fit model for each.
• 2.Separated the analysis pipeline as a NestJS service layer from the API controllers — independently testable and swappable without touching request routing.
• 3.Built report generation as an async pipeline with per-step error boundaries so a failing analysis step degrades gracefully rather than returning a blank result.
• 4.Chose enforced TypeScript contracts between NestJS and Next.js to reduce runtime data shape errors at the API boundary.