Spark

Cold calls are tough. Attention is limited. Spark your opportunities to land your dream client.  

Inspiration

3 CS majors. 100 phone calls. 0 clients landed. 200 more... 0 clients landed. 300 more... 3 clients landed.

Our story is similar to that of thousands of engineer-turned entrepreneurs who have a great idea but lack the outreach expertise. After hundreds of cold calls, we've realized the power of understanding your industry intimately. Now, we're here to optimize your outreach journey by leveraging paralinguistic AI.

What it does

A common misconception among founders is that a failed cold call is a blunder. Spark rejects this notion. We aggregate all your past cold calls to derive insights on interactions that have been successful and unsuccessful for you. Spark recognizes trends in your growth over time and marks timestamps in your cold calls create highlights, instances evoking positive or negative responses. Spark then suggests alternative courses of interaction while operating on a knowledge base of your company powered by a RAG model in order to provide technically informed suggestions.

How we built it

Spark is powered by NextJS and built with TypeScript to provide users with a seamless experience with minimal latency. Our backend and our authentication is based on Supabase. We utilize HumeAI for our cold call audio analysis to provide us with valuable insights on the progression of emotions throughout the call in addition to its transcription and speaker channel diarization capabilities. Our RAG model is powered by OpenAI and allows users to gain personable insights on alternative course of interaction based on heuristics learned from their past conversations and their respective outcomes.

Challenges we ran into

We initially faced trouble with navigating the HumeAI platform as it was our first time utilizing the API. However, using the meticulously crafted API docs allowed us to solve our issues and succeed in utilizing the full power of the API.

Accomplishments that we're proud of

We're proud of creating a product that solves an issue that we previously faced and have heard about extensively from our fellow founders. We are excited to introduce a solution that will enable founders, especially those without marketing or business experience, to expedite the growth of their businesses and by maximizing their potential to land successful cold calls.

What we learned

We delved deep into the core problem and learned about it more in the process of developing Spark. These insights shaped our methodology for the AI pipeline powering Spark.

What's next for Spark

We plan to add live streaming feedback in order to provide users with feedback directly during cold calls using a less computationally intensive model in order to support faster-driven insights. We also hope to reduce latency on the website by supporting better load-balancing practices. 