Inspiration

Freelancers and people early in their careers often have nowhere to practice the client-facing side of the job and understand how business operate in general. This is one of the main reasons experience is required, because companies and freelance clients don't want juniors practicing on their businesses.

What it does

Users take a live client call, clarify scope, upload a logo, get a revision call based on the real submission, and then receive a scorecard plus a gig transcript portfolio page.

How I built it

Built with Next.js, React, TypeScript, and Tailwind. ElevenLabs powers the voice client, Gemini analyzes submissions and scores calls, and MongoDB Atlas stores gig data, transcripts, and result

Challenges Iran into

The hardest part was making the AI client react to the actual uploaded file instead of giving generic feedback. Keeping the voice flow, state machine, and persistence reliable was also a major challenge.

Accomplishments that I'm proud of

I got the full demo path working end to end: briefing, scope, upload, revision, debrief, and transcript portfolio. The revision call can reference the user’s real submission, which is the key moment of the project.

What we learned

A convincing AI product depends on the right context reaching the model at the right time. Clear scope and a strict flow matter more than adding extra features.

What's next for FirstClient

More medium supported, more type of gigs and even harsher clients to train on!

Built With

Share this project:

Updates