The concierge for the calls you keep meaning to make — handled.
+Clutch is a consumer AI concierge that completes the phone calls and life-admin tasks people avoid — booking, cancelling, negotiating, and above all waiting on hold — from a single plain-English request. It’s live in production with eight task types running today; its flagship real-call engine, Holdline, is built, reviewed, deployed behind a flag, and one supervised call from launch.
+ +Booking, cancelling, disputing, chasing a quote, waiting on hold — the life-admin tax. None of it is hard, just tedious and synchronous, so it gets put off, and the friction always wins.
+One inbox, nine task types — Holdline, Sniper, Reservation, Appointment, Bill negotiation, Cancel, Quotes, Waitlist, Errand. Clutch reads the intent, places the calls, and reports back with a receipt.
+AI-cost economics like the point call-agents, task breadth like a human assistant, and a disclosure-first stance — always announces it’s an AI, never records — where rivals have been burned on trust.
+Model: freemium consumer subscription, AI-run cost structure vs. a human-staffed alternative. Figures illustrative.
+A green light to run the first supervised Holdline call and open it to an early-access allow-list. The build risk is behind us — eight task types are live and tested daily. What’s left is validating the one moment that turns Clutch from a task manager into the concierge it’s built to be, and resourcing the team beyond a solo founder-build.
+Give every student honest writing feedback — at the scale of AI.
+Essay Helper is a Microsoft 365 Word add-in that reviews a student’s draft and coaches their citations from inside the document — and it is architecturally incapable of inventing a source. It coaches instead of ghost-writing, meets students where they already write, and it’s built, tested, and shipped today.
+ +One teacher can’t give thirty students timely feedback on argument and evidence. Generic AI fills the gap — but it fabricates citations (misconduct) or just writes the essay (learning lost).
+It flags weak arguments and unsupported claims as comments anchored to the sentence, and coaches in six stances — from Socratic questions to direct, provenance-tagged answers.
+A validation layer drops any citation that doesn’t point to a source the student supplied — so the no-fabrication promise holds even if the model misbehaves. Institutions can approve it because of how it’s built.
+Land free with students; expand to per-seat/per-FTE licenses with admin controls & teacher analytics. Figures illustrative.
+A green light to run a pilot with 2–3 design-partner classes this term, a short runway to a Google Docs client + teacher-facing analytics, and support to court the first institutional buyer. The build risk is behind us — this is now about putting a finished, safe product in front of students and turning real classroom use into the sales story.
+A voice that calls every morning — with a safety net a family can trust.
+Evergreen turns an aging parent’s own phone into a daily voice companion that also screens scam calls before they ring through, backed by a licensed human safety net. The daily call is why she answers; the safety net is why the family pays. It’s built and running in a working prototype today — pre-revenue, with the licensed monitoring partner still to be signed.
+ +57% of Canadians 50+ report feeling lonely, and their adult children often live cities away. Medical-alert pendants wait for a fall; free companion apps wait to be opened. Neither touches the loneliness itself.
+Iris calls every morning on the phone she already owns — nothing to buy or charge. Scam calls are screened in the network before her phone even rings, with a licensed monitoring partner behind an escalation ladder if she doesn’t answer.
+Carrier-native call screening a pure software app can’t copy, plus months of knowing what a good morning sounds like for her. Competitors sell a device, a robot, or a spam blocker alone; Evergreen bundles all three into one call.
+Model: recurring per-parent subscription (Basic / Full / Unlimited), Canada-first via carrier distribution. Figures illustrative.
+A two-week legal-architecture sprint and a signed licensed monitoring partner — then a small team to bring the working Basic product to the first Canadian families. The build risk is behind us: Iris’s morning call, scam screening, and the self-filling CareCircle dashboard already run in prototype. What’s left is the legal wrapper and the licensed safety desk, not the product.
+An academic operating system for the students mainstream tools leave behind — already live in production.
+FocusFlow combines an AI study coach that teaches instead of handing over answers, a full executive-function toolkit, and a safe, verified tutor marketplace in one app — built for the roughly 1 in 5 students who are neurodivergent. It’s not a prototype: 49 data models, 51 screens, and 600+ automated tests, built solo to production grade in about ten weeks. The revenue engine is coded and proven end-to-end against live Stripe (test mode) — one decision from turning on.
+ +Mainstream study tools assume the hard part is content. For a neurodivergent student, the hard part is starting, staying on track, and self-regulating — a layer no point tool owns, so families duct-tape a planner, flashcards, a tutor, and a wellbeing app together.
+An AI coach, executive-function toolkit, full study system, grades & GPA, an accessibility engine, and a safe tutor marketplace — every module writes to the same student model, so the coach gets proactive before a test.
+Accessibility-first architecture and a “teaches, won’t cheat” AI are hard to fast-follow. Near-zero cost to run, and it owns the whole student — so it’s sticky, and the revenue engine is already proven against live Stripe.
+Model: free-forever student core drives adoption; revenue rides tutor subscriptions, a premium family tier, and B2B school licensing — near-zero cost to serve. Figures illustrative.
+A green light to run a small student pilot this term (15–30 real students), then flip the marketplace live. The build risk is behind us — this is now about proving retention with real students and turning a revenue engine already proven in test mode into real dollars.
+Turn empty appointment chairs back into revenue — automatically.
+No-Show Killer is a deployed SMS platform for appointment-based businesses. The moment a client cancels, it races down the waitlist by text and refills the slot in under a minute — and predicts, reminds, and auto-nudges to stop no-shows before they happen. It’s built, tested, live, and one carrier-registration step from its first paying customer.
+ +A no-show is an hour that can never be sold again — 10–15% of revenue at a typical service SMB. The manual fix (calling down a paper waitlist) can’t move in the sub-minute window that matters.
+Cancellation → ranked waitlist cascade → first “YES” books the chair, atomically. Plus AI risk-scoring, confirmations, auto-pilot nudges, and a morning owner brief — all over plain SMS.
+It pays for itself the first time it refills one chair. One product serves salons, dental, physio, gyms, restaurants — and it’s AI-native and genuinely delightful, so owners keep it on.
+Model: recurring per-location SaaS, low cost to serve, expandable to multi-location & tiers. Figures illustrative.
+A green light to finish the A2P registration and recruit 3–5 design-partner businesses. The build risk is behind us — this is now about putting a finished product in front of businesses that lose money every week without it, and turning real recovered dollars into the sales story.
+