Projects & Ideas

Just came up with this. **Phone It In.** A tiny button at the top of every long web form. You click it, it asks for your mobile, it calls you. The voice agent walks you through every question on the form, you answer out loud, and at the end it says "Thanks — form submitted" and hangs up.

For elderly people. For anyone with bad eyesight. For drivers. For the lazy. For anyone who'd rather talk than type. The form already has the field structure — name, email, address, phone, dropdowns, whatever. The agent just reads the HTML, throws the question set at VAPI, captures the answers in a phone call, formats them, and submits.

Best part: we may not even need the website's permission. Make it a **Chrome extension**. Drops the button in via DOM injection, scrapes the form's field structure, dials VAPI on click. Universal — works on every form on the web that the user can see. The site never knows. The site never has to do anything.

This is a $99/year prosumer extension at minimum. Or B2B sold to insurance/healthcare/financial companies whose customers abandon forms at 60% completion.

Open question: where does it break? Captchas, multi-step flows, dynamic JS forms. But for the long boring static-field forms (DMV, government, insurance enrollment, doctor intake, school registration), this is dead simple.

What if you didn't A/B test — you March Madness'd it? Spin up 32 versions of a landing page with Claude doing the design variations, then run them bracket-style. Round of 32, Sweet 16, Elite 8, Final Four, championship. Winner moves on, loser dies.

The questions I can't stop chewing on: what confidence percentage do I trust before killing a page? How many impressions per matchup to hit that confidence? A traditional A/B test wants 95% and thousands of conversions per variant — that's brutal for 32 pages. Do I lower the bar for early rounds (80% at 500 visits) and raise it for the finals (95% at 5,000)? Or do I just trust the agent's design intuition plus a lighter statistical gate?

If this works it's not a test — it's an operating system. Every new product launch goes through the tournament. Champion becomes the canonical page. No opinions. No 'I think this headline is better.' Just reps and math. Feels like the same bracket logic I was drawing up for Emailnado's fleet. Maybe that's the pattern — tournaments as the evaluation layer for everything agents generate.

**Update — the bigger play:** we warehouse the winners. Every bracket champion gets stored in a brain of what works. Not just the page — the components. Which headline pattern beat which. Which CTA shape won when the audience was CFOs vs. marketers. Which hero layout converts higher on mobile. Over time the warehouse stops being a graveyard of past experiments and becomes a CRO knowledge base that feeds the next bracket's Round of 32. The agent doesn't start from zero — it starts from 'here are the 32 patterns most likely to win against this audience, go generate variations.'

That's the compound. Every tournament makes the next one smarter. The moat isn't the design — it's the warehouse.

What if we built a marketing platform that forces you to think like a scientist? Every campaign starts with a hypothesis — 'I believe [audience] will respond to [message] because [reason].' Then you test it. AI runs the experiment — A/B variants, send patterns, audience segments. You get real outcomes, not vanity metrics. Then the AI helps you adjust the hypothesis and run the next cycle.

This is basically what Emailnado's tournament bracket already does, but the framing changes everything. Most marketers throw spaghetti. Scientists iterate. The flywheel isn't 'send more email' — it's 'get smarter every cycle.' Hypothesis → Test → Outcome → Adjust. Repeat forever. AI doesn't replace the marketer's brain — it accelerates the scientific method so you can run 10 experiments in the time it used to take to run one.

What if there was a lightweight web app that helped you achieve fluency in personal development the same way immersion teaches language? Every day you get: a snippet to read, a quote or two to memorize, and a small journal prompt. Your responses go straight into your Second Brain. Over time, the neurons rewire. You don't study success — you become fluent in it.

Could be called Fluency, Personal Fluency, or Success Fluency. Draws from the masters: Ziglar, Tracy, Sharma, Robbins, Rohn, Covey, Hill, Clear, McKeown, Collins, Drucker. Not a course you finish — a practice you maintain. Like Duolingo for the language of leadership.

Tech: lightweight Next.js app, Supabase for journal entries, connects to Second Brain vault. Could even be AI-powered — Claude picks the daily content based on what you've been thinking about lately.

What if marketing ran like a lab instead of a casino? Every campaign starts with a hypothesis. You test it. You measure the outcome. You adjust. And then you do it again — faster, with AI enhancing every loop.

Hypothesis → Test → Outcome → Adjust → Repeat. That's the scientific method. But nobody in marketing actually runs it that way. They run campaigns, look at a dashboard two weeks later, argue about attribution in a meeting, and then do the same thing next quarter.

What if there was a platform that enforced the loop? AI generates the hypothesis based on your historical data. It designs the test. It measures the outcome in real time. It recommends the adjustment. And the flywheel spins again — each cycle smarter than the last. No more gut-feel marketing. No more "we think this worked." Just: here's what we hypothesized, here's what happened, here's what we're doing next.

What if you never had to log into Apollo, ZoomInfo, or any of those confusing CRM enrichment platforms again? Just talk to Leadnado. Tell it who you want to reach and it goes and gets the data for you.

The play is simple: wrap the Apollo API, arbitrage the cost per lead, add a margin, and deliver the results via email. No dashboard. No login. You just say "I need 500 marketing directors in fintech" and Leadnado runs the search, emails you back with a summary, a link to buy the list, and — here's where it gets interesting — a link to have Emailnado or Salesnado sell to them on your behalf.

The tagline: "Stop logging in to confusing CRM enrichment platforms. Just ask Leadnado to go get it for you."

Still working out the economics and how Emailnado fits in. But the idea is a full loop: Leadnado finds them, Emailnado reaches them, Salesnado closes them. All agentic. All via email. No UI anywhere.

I have 4 monitors on my desk right now. Terminal on one, browser on another, Claude Code on a third, and whatever I'm referencing on the fourth. And honestly? I could probably get away with 2.

But here's the thing — vibe coding isn't about screen real estate. It's about flow state. The monitors aren't for multitasking, they're for not breaking context. Every cmd-tab is a tiny interruption. Every window shuffle is a micro-decision your brain didn't need to make.

Quick LinkedIn hitter: "How many monitors does one need to be a vibe coder? I have 4. How many do you have?" Simple engagement post. Let people argue about it.