Inspiration
Payroll and finance tools are fragmented. Employers get processing tools but limited predictive guidance, while employees get little clarity on pay, benefits, and planning. Revo was inspired by building one system that serves both sides with practical AI decision support.
What it does
Revo is a dual-portal AI payroll and finance platform.
Employer portal:
Cash-flow forecasting and liquidity risk alerts CFO-style recommendations with human approval Payroll visibility and what-if simulations Employee portal:
Paycheck and deduction clarity Budget and long-term planning support Benefits guidance and Digital Twin voice assistant It is privacy-first: AI operates on computed metrics/signals instead of unrestricted raw sensitive records.
How we built it
Frontend: Next.js with a unified design system and role-based portals. Backend: FastAPI services with modular endpoints for payroll, simulation, and AI orchestration. AI architecture: Specialized agents (cash flow, CFO, benefits, budgeting, digital twin) connected through an orchestrator. Data layer: Synthetic raw financial/payroll datasets converted into metric CSVs for secure model input. Voice: ElevenLabs conversational integration with context injection from orchestrated agent outputs.
Challenges we ran into
Keeping employer and employee contexts synchronized when payroll changes occur. Making voice agents reliably start across token/WebRTC and websocket fallback paths. Handling runtime edge cases from partially missing data in dynamic UI cards/tables. Maintaining UI consistency while iterating rapidly across many pages. Preventing secrets/agent IDs from leaking during fast iteration.
Accomplishments that we’re proud of
Built a functioning dual-sided AI finance experience, not just static dashboards. Implemented explainable recommendations with simulate-before-approve workflow. Integrated ElevenLabs voice agents into both employer and employee experiences. Established a cleaner, premium, consistent blue/black product identity. Improved security hygiene by moving IDs to env configuration and tightening git ignore policy.
What we learned
Trust and clarity are as important as model quality in fintech UX. Multi-agent orchestration provides more product value than isolated single-agent answers. “Simple, actionable, explainable” wins over feature-heavy complexity. Security guardrails (env management, no hardcoded identifiers) must be part of the core workflow.
What’s next for Revo
Production-grade auth, tenant isolation, and role-based policy enforcement. Stronger scenario modeling with confidence bands and risk decomposition. Proactive alerts (payroll stress, invoice delay risk, benefit optimization opportunities). Deeper integration adapters for payroll/accounting ecosystems. Enterprise audit timeline and approval analytics for operational trust and compliance.
Built With
- elevenlabs
- fastapi
- gemini
- node.js
- python
- react
- react-native
Log in or sign up for Devpost to join the conversation.