Workline AI

• 

  the main ui

Inspiration

When I was working on ideating a hobby project, I ran into multiple problems in building the complete project; even with AI, it took lots of prompts, refining, and time wastage to finally complete the project. So I built a platform that bridges this gap for engineers/hobbyists to completely build their ideas.

What it does

It is a multi-model, self-orchestrating platform that basically completes the whole work from ideation to implementation, giving the builder complete freedom to modify their idea based on the final output

How we built it

We built it using Google ADK, Amazon Bedrock, AWS, Lambda, SurrealDB, QdrantDB, Pydantic, Tavily, LlamaIndex, LangChain, Bindu, x402, Gemini API, Paperbanana, Claude API, Brevo.

Challenges we ran into

We ran into challenges like multiple Agent Coordination, scraping the data without getting banned, researching and organising the components, and making a BOM and connection lists.

Accomplishments that we're proud of

We are proud of building the orchestration layer, external API nodes, DB integration, and agentic Coordination.

What we learned

We learnt about multi-agentic conversations, different ways of embedding data into vector databases, and API cost optimisation.

What's next for Workline AI

Upgrading the code generation for iot, component selection efficiency, BOM report generation, Research paper refining and SDK development

Built With

• amazon-web-services
• claude
• gemini
• googleadk
• graphdb
• lambda
• numpy
• pydantic
• tavily
• vectordb