BrandOS

Inspiration

As a Senior Marketing Analyst, I realized we were optimizing primarily for search while drowning in metrics that lacked strategy. Traditional tools tell you what your competitors are doing, but not how to beat them in the AI age. I built BrandOS to be the first Strategic Intelligence Layer—bridging the gap between raw data and creative application, so brands can stop guessing and start dominating the future of Answer Engines.

What it does

BrandOS solves the strategy gap in modern marketing. While other tools just give you metrics, BrandOS gives you a brain. It acts as an always-on Chief Strategy Officer that deeply understands who you are, detects how the world (and AI) sees you, and gives you the exact creative assets and playbooks to close that gap—in seconds, not weeks.

How we built it

BrandOS isn't just a wrapper; it's a multi-stage reasoning engine. Built on Streamlit for rapid UI and backed by Python, it uses a unique two-stage architecture:

Fast Extraction Layer: Uses Gemini 3 Flash to instantly parse HTML, extract design tokens, and structure factual brand data in real-time. Deep Reasoning Layer: Uses Gemini 3 Pro (the 'Chief Strategy Officer') to process that data, inferring complex buyer psychology, competitive wedges, and strategic gaps that standard models miss. All data is persisted in a Supabase (PostgreSQL) database for long-term memory, ensuring BrandOS learns and adapts alongside the brand.

Challenges we ran into

1. Getting the AI model to be creative with strategy but strict with data structure and avoid any sort of hallucinations.
2. Balancing the ball between deep reasoning and speed (Gemini 3 Pro vs Gemini 3 Flash).
3. Making sure the model did not throw generic marketing fluff (zero context and relevance).

Accomplishments that we're proud of

1. Building a thinking operating system, not Just a wrapper with multistage reasoning architecture.
2. AEO Engine - How LLMs perceive a brand, revealing visibility gaps that traditional SEO tools completely miss.
3. Visual DNA - Scans raw HTML/CSS to extract precise Design Tokens (hex codes, fonts, vibes).

What's next for BrandOS

Community Analytics - Brand performance across communities (Reddit, GitHub). Ability to segment threads/post by sentiment, feature requests, support and potential leads/prospects.

Built With

• altair
• beautiful-soup
• gemini3
• pandas
• playwright
• plotly
• python
• reportlab
• sql
• sqlalchemy
• streamlit
• supabase