RelayIQ

RelayIQ automates employee transitions end-to-end by capturing knowledge, generating personalized training, and making handoffs seamless and stress-free.

Comment

Inspiration

While working at a small business, a developer left unexpectedly and it took us weeks to onboard the replacement. Someone always had to be available to answer questions and clear doubts, which was difficult with limited resources. Most of the knowledge was scattered across past work, and there was no reliable way to transfer it quickly or completely.

What it does

RelayIQ captures real working knowledge and turns it into focused, role-specific training so the next person doesn’t have to start from zero. It includes a Rovo Agent powered by LLMs that understands which training module the trainee is viewing, fetches the most relevant embeddings, and delivers clear, contextual answers whenever they get stuck.

How I built it

I built RelayIQ by fetching work data, converting it into embeddings, and using semantic search to identify what actually matters. Large language models then use this filtered knowledge to generate structured training modules. The system is designed to stay flexible, with human control built into how content is generated and refined.

Challenges I ran into

Handling large volumes of data caused timeout issues early on. I solved this by breaking the data into smaller chunks and processing them in parallel using Atlassian Forge. Another challenge was keeping knowledge current while employees continued working, which led to building a daily sync feature that continuously updates embeddings so training always reflects the latest information.

Accomplishments that I'm proud of

I’m proud that I was able to take what already existed scattered work and incomplete knowledge and turn it into something genuinely useful. RelayIQ maximizes existing data and turns it into practical training that reduces onboarding time and dependency on others.

What I learned

I learned that onboarding slows teams down not because people don’t want to help, but because knowledge isn’t organized. Context matters more than volume, and AI works best when it removes friction instead of adding complexity.

What's next for RelayIQ

I plan to improve both the input and output sides of RelayIQ. On the input side, I want to integrate more third-party business tools that companies already use, allowing RelayIQ to capture a broader and more complete view of employee knowledge. On the output side, I plan to introduce richer training formats like interactive content, video, and audio. RelayIQ is evolving into a long-term knowledge continuity platform as AI capabilities continue to grow.

Built With

Share this project:

Updates