Secula

Runtime Sessions
Studio to set up swarm
Chat interface

Inspiration Wall Street has a monopoly on speed, data, and coordination. Everyday retail traders are playing a game they are mathematically designed to lose. We were inspired to level the playing field by bringing institutional grade quantitative analysis directly to the common investor. We wanted to give everyday people the power of a tireless and highly specialized quantitative trading firm working around the clock.

What it does Secula is a fully autonomous trading orchestrator. Users connect their brokerage APIs and define their custom orchestrations. Instead of staring at charts, users build their own AI hedge fund by setting specific agent personas and assembling a dedicated research team. You might assign one agent to scrape global financial news, another to run sentiment analysis on social media, and a third to crunch the quantitative market data. These agents work together in perfect synchronization to analyze the markets and execute trades autonomously while strictly following the risk guardrails you put in place.

How we built it We built the core orchestration engine using Python, LangChain, FASTAPI, Supabase, OpenAI, Ollama, LangGraph, Node, and Playright to manage the complex multi agent network. The data pipeline handles web scraping and ingests market data which is then fed into our machine learning sentiment analysis models. We developed a Cron synchronized execution system to ensure all agents coordinate their research and act on the data at precise intervals.

Challenges we ran into Ensuring perfect synchronization between the different agent personas was a major hurdle. We had to handle rate limits and format inconsistencies across various news and financial APIs. Additionally, designing a robust guardrail system that could perfectly interpret user defined risk parameters and safely halt any rogue agent trades required significant architectural planning and rigorous testing.

Accomplishments that we're proud of We successfully orchestrated a multi agent system where different personas can independently research and then collaborate seamlessly to execute a unified trading strategy. We are also incredibly proud of building an intuitive flow where users can easily define their research team and set orchestrations without needing a background in quantitative finance.