Boast: Turning Cold Outbound into Inbound 🚀

Inspiration

Every small startup founder knows the hardest part of sales isn’t the product — it’s the pitch. Cold emails, ignored LinkedIn messages, and missed calls are draining. Founders can’t scale themselves, but scaling genuine human connection felt impossible — until now.

We built Boast to turn every founder into a scalable, 24/7 sales team. The idea came from the frustration of trying to reach investors and customers who never had time to hop on a call. What if your pitch could meet people instantly — even when you’re asleep?

What it does

Boast is an AI-powered sales tool that turns cold outbound into inbound.

It lets founders and sales teams create hyper-real, real-time avatars that host interactive LiveKit meetings with leads and investors. These avatars deliver personalized sales pitches, handle objections, and answer questions — all with your exact tone, script, and boundaries.

Instead of sending static decks or cold emails, you share a Boast link. Prospects join a live AI meeting where they talk to your avatar, experience your pitch, and leave their information — converting cold traffic into warm, qualified leads automatically.

How we built it

We built Boast over a hackathon weekend using:

Next.js – for the app interface and server-side rendering
Supabase – for user authentication, data sync, and analytics storage
LiveKit – for real-time audio/video and avatar interactions
Cursor – to accelerate AI-assisted code generation
Vercel – for fast CI/CD and edge deployments

Our focus was on creating seamless, low-latency meetings between avatars and users — powered by AI-driven speech, movement, and response generation. Supabase handled meeting data, engagement logs, and pitch performance analytics. Everything runs on Vercel’s global edge network for sub-second response times.

Challenges we ran into

Real-time latency – Balancing high-fidelity avatar rendering with <200ms response time across global meetings was tough. LiveKit helped, but optimizing peer connections and stream priorities required deep debugging.
Avatar realism – Creating avatars that looked human but didn’t cross into uncanny territory required tuning animation and emotion models carefully.
Sales truthfulness – The avatars needed to stay on script while still sounding conversational. We implemented “guardrails” to keep every response factual and aligned with pre-approved sales decks.
Session handoffs – When an avatar meeting generated a lead, automatically syncing that data to Supabase and CRM tools without breaking session flow was tricky.
Cold → warm conversion – Measuring how effectively avatars turned uninterested visitors into engaged prospects involved designing custom event analytics for “attention,” “engagement,” and “intent signals.”

Accomplishments that we're proud of

Delivered a working hyper-real avatar demo using LiveKit within 48 hours.
Achieved under 200ms interaction latency globally via optimized signaling servers.
Built a founder-facing dashboard showing meeting replays, engagement metrics, and lead analytics.
Got immediate traction — several hackathon judges used Boast to pitch their own projects.

What we learned

We learned that sales doesn’t have to start cold — it can start with presence. Founders don’t need more tools; they need scalable authenticity.

Technically, we learned about real-time media pipelines, WebRTC debugging, guardrailed AI dialogue, and how to design systems that feel alive without losing control.
But the biggest takeaway: selling is storytelling — and AI can finally help founders tell theirs at scale.

What's next for Boast

We’re now building toward:

CRM integrations (HubSpot, Salesforce) for seamless data flow
Multi-agent meetings for group demos
Voice cloning and emotional tuning for more human-like delivery
Personalized follow-ups via AI email after each meeting
Automated lead scoring based on engagement analytics

Our vision: transform Boast from a hackathon prototype into a full AI-powered sales companion that makes every founder their own best salesperson — even while they sleep.