-
-
Axiom99
Inspiration
Modern financial engineering, regulatory compliance, and software execution pipelines are crippled by fragmentation and slow processing times. We were inspired to build Axiom99—a universal, multi-agent autonomous platform designed to streamline end-to-end development lifecycles, risk auditing, and real-time portfolio rebalancing. Driven by the goal of empowering communities and developers with automated intelligence, we set out to create a 99-agent swarm architecture that turns complex workflows into parallel, seamless execution loops.
What it does
Axiom99 operates as an autonomous multi-agent engine that automates complex technical and financial tasks. It continuously audits regulatory compliance, evaluates risk profiles, and executes adaptive portfolio rebalancing.
The platform evaluates real-time risk scores R(s) across parallel agent execution streams using an aggregated weighted decay function:
R(s) = \sum_{i=1}^{99} w_i \cdot f_i(s_i) \cdot e^{-\lambda t}
Where w_i represents agent weight, f_i(s_i) is the individual risk output, and \lambda is the temporal decay factor.
How we built it Multi-Agent Engine: Architected a swarm of 99 specialized agents coordinating execution across sub-tasks. Frontend & Real-Time Stack: Developed with Next.js for an interactive user interface alongside Supabase and Firebase for low-latency state persistence and real-time data sync. Predictive Analytics & ML: Integrated XGBoost for real-time predictive risk scoring and system anomaly detection. Deployment & CI/CD: Connected continuous integration and deployment pipelines using GitHub and Vercel for automated platform updates. Challenges we ran into Agent Coordination & State Synchronization: Managing state consistency across a high-throughput swarm of 99 autonomous agents introduced potential race conditions. We resolved this by implementing centralized event orchestration via Supabase and Firebase real-time listeners. Low-Latency Risk Inference: Running predictive risk models rapidly across streaming data streams required optimizing our XGBoost feature pipelines to achieve sub-second execution. Accomplishments that we're proud of Built a fully functional 99-agent autonomous architecture capable of handling complex fintech and software creation workflows end-to-end. Successfully integrated real-time ML risk predictions alongside interactive mathematical verification. Implemented a continuous CI/CD pipeline hosted on Vercel with direct GitHub automation. What we learned Effective design patterns for multi-agent swarm orchestration and load distribution. Strategies for combining predictive machine learning models with real-time web databases like Firebase and Supabase. Modern approaches to automating compliance, risk auditing, and software build pipelines. What's next for Axiom99 Expanded Fintech Integrations: Integrate direct API hooks for dynamic liquidity protocols and algorithmic trading environments. Agent Personalization: Allow developers to define custom rules, agent behaviors, and verification constraints via standard configuration files. Community Deployment: Scale Axiom99 as an open-source tool to support community developers and hackathon projects worldwide.
Log in or sign up for Devpost to join the conversation.