https://www.loom.com/share/2f99b4e2030046588706a7e00599a9a3

Inspiration Selling high-ticket services on sales calls is the single highest-leverage skill for online fitness coaches and agency owners, yet it has no dedicated training product. Coaches and agency owners learn to sell by losing real deals. Every botched objection, every awkward price reveal, every "let me think about it" that goes unhandled costs real revenue and erodes confidence. We built Leetclose because we saw three things converge at the right time: voice AI finally reaching human-level naturalness, a massive underserved market of 300K+ online coaches with zero structured sales process, and a proven retention model from Duolingo that nobody had applied to sales training.

What it does Leetclose is an AI-powered sales call practice simulator for anyone who closes high-ticket offers on calls. It places users into realistic, voice-based simulated sales calls against AI prospects where they practice the full lifecycle of a high-ticket conversation: building rapport, discovering pain, handling objections, pitching the offer, revealing price, and asking for the sale. After each call, users receive instant scoring across six dimensions (rapport, discovery, pain extraction, objection handling, offer clarity, and closing strength) along with transcript review and targeted improvement guidance. The product draws from three proven models: Duolingo's habit-forming streaks and XP, competitive gaming's rank progression and leaderboards, and consultative (NEPQ-style) sales methodology where the prospect self-discovers their problem rather than being hard-pitched.

How we built it We started with a clear PRD defining our two target personas (fitness coaches and agency owners), mapped the full problem space, and prioritized features into P0/P1/P2 tiers. For the hackathon MVP, we focused on the core loop: sign up, start a practice call, speak with an AI prospect in real time, and receive a post-call performance score. We used real-time voice AI to power the prospect engine, maintaining persona consistency through structured persona cards that hold backstory, tone, goals, objections, urgency, and buying context. The AI prospect responds dynamically rather than following a rigid script, generating realistic objections based on timing and user performance, and adjusting resistance depending on call quality.

Challenges we ran into Keeping AI response latency under two seconds was critical since anything above that breaks the immersion of a live sales conversation. Maintaining persona consistency over longer calls was another challenge, which we solved by giving each prospect a fixed persona card that the AI references throughout the call to avoid drift. We also had to balance making the AI prospect realistic enough to create genuine practice value without making it so difficult that new users got discouraged. Building a scoring system that accurately captures nuanced sales behaviors like tonality, timing, and emotional intelligence rather than just keyword matching was another significant hurdle.

Accomplishments that we're proud of We validated the full end-to-end core loop within a single hackathon sprint: a user can sign up, enter a realistic voice-based sales call with a dynamic AI prospect, and walk away with actionable scoring and feedback. The simulation genuinely feels like a real sales conversation, not a chatbot interaction. We also designed a complete product roadmap that takes Leetclose from hackathon MVP through to a scalable platform covering multiple sales verticals. The positioning at the intersection of Duolingo, competitive gaming, and consultative sales methodology is something we believe is truly novel in this space.

What we learned We learned that the gap between passive sales training (scripts, courses, videos) and active realistic practice is even larger than we expected. The technology to close that gap only became viable in the last year with real-time conversational AI reaching sub-second latency. We also learned that ethical, consultative selling (asking better questions, listening deeply, leading with empathy) is far more trainable through simulation than through theory alone. Building for a specific niche first (fitness coaches) rather than trying to serve everyone gave us sharper product decisions and clearer feedback loops.

What's next for LeetClose Our roadmap spans four milestones. Over months one and two, we are expanding into a more complete product with onboarding personalization, a broader prospect persona library, stronger scoring logic, and trial-to-paid billing to validate product-market fit. Months three through six focus on retention: gamification, dedicated training modes (objection-only drills, closing practice, price-reveal drills), and daily engagement loops. By months six through twelve, we plan to expand beyond fitness and agency into additional call-based sales verticals like real estate, consulting, insurance, and SaaS, while introducing team and enterprise features. The long-term vision is to turn Leetclose from a niche simulator into the go-to platform for training call-based selling skills across industries.

Built With

Share this project:

Updates