PIP-SMB

You AI executive assistant, helping you book more calls, close more deals, and save more time.

Comment

Try it live -- no sign-up needed: proxyintelligentpartner.com -- log in with demo@proxyintelligentpartner.com / XPRIZE2026Demo! to explore a live, populated PIP workspace (leads, pipeline, a closed deal, tasks) as a business owner would see it.

Or call the AI receptionist directly: +1 (639) 739-6688 -- PIP answers, understands the request, and books it straight onto the calendar in real time. Same live workspace as above.

Inspiration

Every small business owner I know is doing the same thing: answering their own phone, chasing their own leads, running the business at night because the day got eaten by admin. I saw it firsthand coaching at a boxing gym in Regina, Saskatchewan, where the owner was doing sales, scheduling, and lead follow-up entirely by hand on top of actually running classes. I built PIP because that shouldn't be the job. Small businesses like gyms, cleaning companies, and solo consultants are too small to hire a receptionist or a sales-ops team, and too busy to build one themselves -- and they're the businesses most likely to lose a customer simply because nobody picked up the phone.

What it does

PIP is an AI-native operations platform that runs the front office for small businesses. It answers the phone through an AI voice agent that understands the caller's intent and books the appointment directly onto the business's calendar -- zero human involvement. It follows up on leads, tracks a real sales pipeline, and logs every call outcome automatically. "Road Trip" mode takes it further: a fully voice-driven, hands-free workflow where PIP briefs the owner on their next lead out loud, dials on a spoken command, logs the outcome, drafts a follow-up text, and sends it on voice approval -- an entire sales motion run without touching a screen, built for owners who are in the field, not at a desk. "Ask PIP" is a conversational co-pilot available anytime -- ask it about your schedule or who to follow up with, or tell it to log a task, and it just does it. Every business type ships with its own defaults: a gym owner sees membership and trial-conversion language, a cleaning company sees job value and quote win rate, a consultant sees deal value and close ratio.

How we built it

Next.js on Vercel, with Supabase for the database and auth. The Gemini API is the reasoning engine behind Ask PIP's conversational responses, lead scoring, and call summarization. Vapi.ai plus Gemini Live power the inbound voice agent; ElevenLabs handles PIP's spoken voice; Twilio sends SMS follow-ups; the Google Calendar API keeps appointments synced two-way in real time; Stripe runs billing. I'm a solo, non-technical founder -- I directed an AI coding agent (Claude) to build, review, and ship the actual product, so the engineering itself was largely AI-executed under my direction, while every business decision, priority call, and the actual sales conversation that turned into Crown Boxing's signature were entirely mine.

Challenges we ran into

I started this build roughly ten days behind my own plan, working on it alongside a full-time coaching job. Getting the AI voice agent to reliably understand real callers -- not just clean demo audio -- took several iterations. Mobile voice input needed real-device testing across browsers before I trusted it enough to demo. The hardest challenge wasn't technical: it was bringing a real product to a gym I already coach at and asking them to trust it with real customer calls while it was still in beta.

Accomplishments that we're proud of

Nothing in this submission is staged. As of submission, PIP's production database shows 64 timestamped agent-execution log entries, 92 real AI conversation turns, and 42 logged outbound calls with a 62% connection rate -- all viewable inside the product on the Agent Logs and Stats pages. Crown Boxing, the Regina gym I coach at, is PIP's first paying customer: a real $50.00 CAD Founding Member subscription, charged and settled through Stripe on August 11, 2026. Five real users now run their day-to-day front office through PIP -- coaches see their own schedules, the owner sees every lead and deal, and the AI receptionist handles their calls.

What we learned

Building the hands-free Road Trip workflow taught me how much a voice interface needs deliberate checkpoints -- a spoken confirmation before anything sends -- to feel trustworthy instead of reckless. Directing an AI coding agent at this scope taught me where velocity helps and where it doesn't: fast for implementation, useless without a human holding the actual product judgment. And bringing PIP to a real customer, rather than only testing it myself, changed what I prioritized -- real usage surfaced problems, like calendar sync edge cases, that my own testing never would have found.

What's next for PIP-SMB

Revenue before August was $0 -- this only became a real paid business in the final stretch of the hackathon period, and I'm stating that plainly rather than smoothing it over. What's shipped is not a finished product -- the analytics are still thin with one paying customer, and there's real polish left to do. But the inbound AI receptionist, the hands-free Road Trip workflow, and the live Stripe subscription are all running in production today, not staged for a submission. Beyond the hackathon, the plan is a real, long-term small business software company: broader templates for more business types, deeper AI reasoning like predictive lead scoring, team features as customers add staff, and continued rollout to new gyms, cleaning companies, and consulting practices. This contest was the forcing function that got PIP to its first paying customer ten days late -- it was never meant to be the finish line.

Built With

Share this project:

Updates