The HackNiche Crucible & The "Free Lunch" Reboot
StratLens 360 was born in the trenches of the HackNiche 4.0 hackathon. My friend and I entered as a two-person team, fully aware we were going up against massive 4-person squads. We picked the "Oracle AI Restaurant Manager" problem statement because we thought we had the stack figured out.Two hours in, we were completely cooked. The scraping architecture was frying our brains, and the implementation felt impossible. We legitimately decided to just grab the free lunch and leave. But sitting there eating, we rebooted our plan from scratch. We decided to stay, took shifts grabbing one hour of sleep each, and ground out the next 24 hours. We ended up taking 3rd place overall.
Taming the Multimodal Chaos
The vision for StratLens 360 was to stop relying purely on 5-star ratings and tap into raw, unbiased opinions: short-form video and Reddit.To process an Instagram Reel, the backend downloads the video and runs it through GCP Computer Vision to extract aesthetic visual cues and color palettes. Simultaneously, we pull audio transcripts, captions, and comments.
The hardest part was normalizing this chaos into a unified BI framework (SWOT, 4P). We built a weighting system. Personally, I knew Reddit data was the most "legit" and raw, so it deserved the highest weight. However, to comply with the specific hackathon grading rubrics, we had to dynamically rig the weights to prioritize Google Reviews at the top.
The Benne Demo
You can build all the data pipelines you want, but a BI tool is useless if it only states the obvious. For our live hackathon demo, we ran the engine on Benne, Bandra's insanely hyped Darshini-style dosa spot.The AI didn't just summarize reviews; it acted like a McKinsey consultant. It identified that Benne's extreme waiting times were generating massive FOMO, but causing severe customer fatigue. The model's actionable suggestion? Install physical shades outside for the queue. It also warned that Benne was currently riding a massive trend wave, and they needed to innovate beyond just the "benne dosa" to survive when the hype died down.
Extreme Usability: The WhatsApp Agent
Graphs and metric dashboards look cool to developers, but they are incredibly overwhelming for a rookie restaurant owner who just wants to know how their business is doing.Our ultimate goal was extreme usability. We eradicated the learning curve by exposing our APIs to Twilio and ChatGPT. Instead of forcing owners to learn a new UI, we built a chat agent. An owner could literally open WhatsApp and text, "How are my sales?" or "What's the sentiment today?" The LLM integration allowed the bot to adapt to the owner's tone and explain complex metrics in plain, personalized language.
