Salesguy.exe | Devpost

Problem

People have different preferences when it comes to:

How much info they need to make a decision
How risk averse they are
How they like to communicate

...and more.

A one size-fits-all approach to sales makes the user less likely to feel understood and catered to.

Solution

Analyze the user's personality and cater your communication to the customer's preferences, making it more likely that the user buys and sticks around in the long run.

Inspiration

We have an existing CRM that helps retention heavy companies deliver better customer support and customer success.

That tool also uses a personality model but the data entry is quite "dumb" as it's just a form.

We were brainstorming ways to detect this automatically and other forms or data input like calls/video instead of just text.

I have a friend at a startup that heavily relies on ad-hoc sales calls that seem to be a good fit for this as you don't have much data on who is calling upfront like you do with scheduled meetings.

So we thought we'd use this Hackathon to explore the intersection of both ideas and build a prototype.

What it does

We help salespeople increase their conversions (by up to 300%) by analyzing the customers' personality.

We then show the salesperson recommendations that they can implement to steer the conversation to the preferences of the customer.

How we built it

We built a live chat SaaS that you can embed into a website with an iframe similar to Intercom and a full screen "dashboard" for the operator.

To analyze the customer's personality, we send the sentences they speak on the call to an API that accepts text inputs from the user to identify their personality and also get their LinkedIn and Twitter handles from Clearbit to seed it with more data.

The profile is then updated as the call progresses, but can start with a solid foundation.

We use Agora.io for the live call and AssemblyAI for the real-time transcription.

Challenges we ran into

At first, we only wanted to use the call transcription for this, but it produced results that were too generic or took too long to accumulate because we were lacking good data.

We introduced Clearbit to solve this and build on top of a foundation of data with the LinkedIn and Twitter profiles.

We also had some issues with API limits :)

Accomplishments that we're proud of

That we challenged ourselves to use remix instead of our normal stack next.js
That the realtime transcription is performant
That we built it in public on Twitter

What we learned

How remix works
How Assembly realtime transcripts work
That making videos takes a long time 😱

What's next for Salesguy.exe

This is a good candidate for a MicroSaaS, but for distribution it might make more sense to embed this in existing tools like a zoom extension or embedded into intercom.

We will think about the ROI as we already have too many projects 😏