LAKSHMAN AI (Resolve Al)

ResolveAI is an AI-powered agentic support system that understands issues, finds relevant solutions, takes actions, follows policies, and creates tickets—automating support end-to-end.

Comment

Inspiration

ResolveAI was inspired by a common problem faced by organizations: employees often struggle to get quick help for issues related to IT access, HR queries, security concerns, and general support requests. Traditional support systems can be slow because every request may require manual triage, searching through knowledge bases, assigning tickets, and performing repetitive actions. We wanted to build an intelligent system that could understand a user's problem and automatically decide what type of support is needed, which information is relevant, and what action should happen next. That idea led to Resolve AI — an Agentic AI-powered intelligent support and resolution platform

What it does

Resolve AI uses AI agents to understand user issues, categorize them, retrieve relevant information, suggest or perform appropriate actions, route requests to the right department, and create support tickets for faster resolution.**

How we built it

We developed Resolve AI using a modular architecture with separate agents and services. he backend was built using: Node.js Express.js Google Generative AI / Gemini MongoDB REST APIs JavaScript The project was structured around independent AI agents such as: triageAgent.js knowledgeAgent.js actionAgents.js hrAgent.js The agents communicate through the main resolution workflow. The system receives a user query through an API endpoint, processes it through the appropriate agents.

Challenges we ran into

One of our biggest challenges was making the agents work together in a reliable workflow instead of behaving like independent chatbots. We also faced several technical challenges, including: Integrating and configuring the Gemini API. Handling API key and model availability issues. Designing accurate issue categorization. Routing requests to the correct agent. Connecting the knowledge retrieval and action workflows. Creating structured responses for tickets and resolution actions. Managing backend project structure and server configuration.

Debugging these issues taught us the importance of modular architecture, API testing, error handling, and iterative

Accomplishments that we're proud of

We are proud of successfully turning the idea of ResolveAI into a working multi-agent AI prototype. Our system can intelligently understand user issues, route them to the appropriate agent, and provide structured resolutions.

We are especially proud of overcoming technical challenges and building a functional backend workflow from scratch.

What we learned

Building Resolve AI helped us gain practical experience with:

Agentic AI architecture Multi-agent systems Backend API development Intent classification and routing Knowledge retrieval MongoDB integration Debugging and testing Node.js applications

What's next for LAKSHMAN AI (Resolve Al)

We plan to continue improving ResolveAI by adding:

Real-time integrations with enterprise tools. Authentication and role-based access. A more advanced knowledge base using vector search. Human approval workflows for sensitive actions. Analytics and dashboards. More specialized AI agents. Deployment as a scalable SaaS platform.

Our vision is to make ResolveAI an AI-powered autonomous support layer for organizations, helping teams resolve repetitive requests faster while allowing human experts to focus on complex problems.

Built With Node.js Express.js JavaScript MongoDB Google Generative AI Gemini REST APIs Agentic AI Multi-Agent Architecture GitHub

Built With

Share this project:

Updates