Voice-Based AI LinkedIn Lookup Assistant

Inspiration

As omnichannel content and digital experience experts, our customers have long asked for more accessible tools, allowing them to expand on their mission to provide digital equality and inclusion. Voice assistants already provide a hands-free resource for users with physical disabilities, but they have long lacked the nuance and context to be truly effective. With the advancements in AI, we now have the means to customize voice assistants with conversational capabilities that can be applied to work-related applications while reducing the dependency on live operators.

What it does

The AI LinkedIn Lookup Assistant is an easy-to-use voice-based system that enables users to request data from LinkedIn regarding specific people simply by speaking a name. The request can provide a wide range of specific details, from the place of employment to the geographic location.

How we built it

The Node.js app harnesses Vapi to provide the vocal layer and LLM capabilities powered by Amazon Bedrock. Along with tapping Bright Data's MCP, Datadog's Watchdog capabilities provide continuous real-time session monitoring of the application.

Challenges we ran into

MCP is always more challenging than you think, creating some friction. Latency is also a persistent issue with the APIs and returning answers to queries.

Accomplishments that we're proud of

Getting the voice features customized around our specific goals, and knowing it could support a wide range of users.

What we learned

Voice is a sensitive and challenging engagement channel and requires testing and ongoing optimization.

What's next for Voice-Based AI Sales Assistant for LinkedIn Lead Research

As part of our Osirus AI package, we see the application of all three integrations as a long-term features as we expand our ecosystem and target funding for our AI platform.