Never lose flow again
with Danny Newman, On Demand Human
Never lose flow again
Show Notes
Danny Newman built ID345 - Denver's AI clubhouse and vibe coding community - from a biweekly meetup at a coffee shop into a weekly gathering of 300+ builders, a coworking space in a converted warehouse, and an alternating generative media meetup. The community spawned its own product: On Demand Human, an in-flow expert marketplace built to solve the exact problem the community kept surfacing.
The insight was simple and painful: vibe coding gets you 80–90% there, fast. Then something specific breaks. You've tried ChatGPT, Perplexity, YouTube tutorials, re-prompting 47 different ways. The energy leaves your body. The project dies on your laptop. What you actually need is a real person who knows the exact tool you're using - available now, not in three days after an Upwork RFP process.
On Demand Human: In-Flow Expert Help
The product is a real-time marketplace for expert help during vibe coding sessions. You select the tool you're using (Lovable, Bolt, Cursor, etc.), describe the problem, set a maximum payment, and request a helper. The platform matches you in real time with an available expert - probably someone between their own projects or with 30 minutes of downtime - who joins you on Zoom with screen share. Pay per minute. The session lasts 10–20 minutes. You stay in flow.
On the supply side, helpers log in, set their per-minute rate, and get browser or email notifications when requests matching their expertise come in. First to accept gets the session. It's designed to feel like office hours, not a side gig - casual availability with upside, no job-posting maintenance required. GitHub login lets the platform verify expertise from public repos.
The Virtuous Expertise Ladder
Danny's observation from running ID345: the person who was stuck on a Lovable database query six months ago is now the ideal helper for someone stuck on the same problem today. They just solved it. They remember the frustration. They can explain it in the language of someone who isn't a senior engineer.
This creates a self-sustaining supply side. As the community graduates through levels - from first prototype to production app to helping others - expertise flows back into the marketplace naturally. The knowledge ladder is both the supply mechanism and the community glue.
Why Pay-Per-Minute Changes the Behavior
Fiverr and Upwork are fine for planned, async work. They're the wrong tool for an in-flow blocker. The multi-day process of posting, vetting, scoping, and contracting is itself a flow-breaking exercise. By the time you've found someone on Upwork, you've already lost the session.
Pay-per-minute solves three things: it makes the value transparent to the person asking for help ("am I getting 20 minutes of value from this?"); it makes availability casual enough for helpers to treat it like office hours; and it creates no barrier for experts who would never manage a Fiverr profile but will happily field a session while they're waiting for something else. The startup use case Danny calls "sneaky powerful": use it as instant office hours for your entire team - any moment where a 15-minute nudge saves a week stall.
Audience → Community → Product
ID345 is a live proof of Greg Eisenberg's framework: build an audience, let it select into a community around a shared interest, then build a product that serves the community's specific pain. Danny didn't start with On Demand Human. He started with a meetup. The meetup became a community. The community's recurring frustration (hitting walls, losing flow) became the product spec.
The result: On Demand Human launched into a community that already exists, already trusts Danny, already experiences the exact problem. Early distribution, instant validation, no cold-start. The product is a natural extension of the community's identity.
Technical Skills Are Now Table Stakes
When anything can be built with vibe coding tools, technical ability is no longer the differentiating variable. Danny's observation from watching hundreds of builders in the ID345 community: the people who succeed aren't necessarily the strongest engineers. They're the people who figure out marketing, distribution, customer acquisition, and revenue. Sam Altman's "era of the ideas person" is real - but the ideas aren't enough either. What matters now is whether you can get to a customer.
- Audience → Community → Product - Build an audience first, let it select into a community, then build the product that serves the community's specific pain. Launches with distribution, validation, and early adopters already in place. Greg Eisenberg's framework, applied live.
- The Flow-First Product Principle - Design around protecting momentum, not just delivering features. The most valuable intervention isn't the best feature - it's the one that keeps someone from abandoning their project. Measure product value in momentum preserved.
- The Virtuous Expertise Ladder - In peer expertise marketplaces, today's beginner becomes tomorrow's best helper. People who just solved a problem are more empathetic and contextually accurate teachers than senior experts who solved it years ago. Design supply-side mechanisms that capture this naturally.
- On Demand Human - Danny's in-flow expert marketplace: select your tool, set a max payment, get matched in real time with an expert on Zoom. Pay per minute. Stay in flow.
- Lovable - Primary prototyping platform referenced throughout; the tool most users hit the 80–90% wall on before needing On Demand Human.
- Bolt - Vibe coding tool in the ID345 community's toolbox; one of the platforms helpers specialize in.
- Replit - Online IDE mentioned as one of the early tools in the community's exploration.
- Cursor - Category 1 coding assistant; how Danny first shipped an iPhone app over Thanksgiving 2024 and how the community began.
- Claude Code - Ryan's next step after Lovable hits the wall; the graduation path for taking a prototype toward production.
- BrainGrid - Ryan's tool for generating proper prompts to bring into Claude Code; mentioned as a bridge between prototyping and deeper development.
- n8n - Workflow automation platform; mentioned alongside Claude Code as a tool with a steeper learning curve that benefits from expert help.
- Perplexity - One of the tools founders cycle through when stuck before giving up; named as a signal that you've hit the wall and need On Demand Human.
- Flow State (in building) - The mental state of full engagement when you know the next step, and the next, and momentum is building. Characterized by energy, direction, and the feeling that today you will finish this thing. When broken by an unresolved technical blocker, projects don't pause - they die. Protecting flow state is the design principle behind On Demand Human.
- The 80–90% Wall - The point in a vibe coding session where forward progress stalls on a specific, unresolvable blocker: a database query that won't behave, an integration that won't connect, a feature that prompting can't unlock. You've tried ChatGPT, Perplexity, YouTube. Re-prompting produces no new answers. Energy leaves. Project dies. The diagnostic: if you've tried it three different ways and you're on your fourth, you've hit the wall.
- In-Flow Expert Marketplace - A real-time marketplace model where experts are available immediately (like Uber for knowledge workers), matched to specific tool or domain problems, priced per minute. Contrasted with async platforms (Fiverr, Upwork) that require days of posting, vetting, and scoping - themselves flow-breaking activities. Designed to feel like office hours, not a freelance engagement.
- Pay-Per-Minute Pricing - A pricing model for time-based expert services that provides full transparency to both sides, lowers the friction threshold for helpers to make themselves casually available, and aligns incentives toward speed and resolution. Makes the value exchange continuous and visible: both parties know exactly what's being paid and earned per minute of session time.
- The Virtuous Expertise Ladder - The self-reinforcing dynamic in peer expertise marketplaces: people who just solved a problem are the most empathetic and accurate helpers for people currently facing it. They remember the frustration, they speak the same language, and they just graduated to the next level. Creates natural supply-side growth as a community matures.
- Audience → Community → Product - Greg Eisenberg's product development framework: build an audience around a shared interest, let it self-select into a community with shared identity and pain, then build the product that solves the community's specific problem. Produces warm distribution and instant validation at launch. ID345 to On Demand Human is the canonical example.
- Technical Skills as Table Stakes - The post-vibe-coding shift in what defines a successful founder: when AI tools remove technical barriers for everyone, engineering ability stops being a differentiator. The new scarce resource is marketing, distribution, customer acquisition, and revenue generation. The "era of the ideas person" (Sam Altman) is real, but the actual winner is the person who can get the idea in front of a paying customer.
What problem does On Demand Human solve, and why now?
It solves the in-flow blocker problem: you're 80-90% through a vibe coding project, something specific breaks, you've tried ChatGPT, Perplexity, YouTube tutorials, re-prompting 47 different ways - nothing works. The energy leaves your body. The project dies. What you actually need is a real person who knows the exact tool you're using, available right now, not in three days after an Upwork process. The timing is right because vibe coding tools have democratized building to the point where the bottleneck is no longer starting - it's the specific blocker that stops you in the middle.
How does the pay-per-minute model work for both sides?
For the person asking for help: you select your tool, describe the problem, set a maximum payment (say $20 or $50), and request a helper. You're matched in real time on Zoom with screen share - the helper can even take over your screen if needed. You pay by the minute; a typical session runs 10-20 minutes. For helpers: you log in, set your per-minute rate (25 cents, a dollar, three dollars - whatever reflects your value), and turn on browser or email notifications. When a matching request comes in, first to accept gets it. The model is designed to feel like office hours - I've got 30 minutes, let me make myself available and see what comes in - not like managing a freelance portfolio.
What's the virtuous cycle you see in the On Demand Human model?
Someone who was struggling with a Lovable database query six months ago is now the ideal helper for someone stuck on the same thing today. They just solved it, they remember the frustration, they speak the language of someone who isn't a senior engineer, and they can explain the fix in terms that actually land. As the ID345 community graduates - first prototype to production app to helping others - expertise naturally flows back into the marketplace. The knowledge ladder creates the supply side. You don't need to recruit senior engineers; you need a community that's actively leveling up, and the freshly-leveled members are the best teachers.
How did building the ID345 community first make On Demand Human easier to launch?
It solved the cold-start problem on both sides. The community already exists, already trusts Danny, and already experiences the exact problem every week. Early adopters, first helpers, first customers - they're all the same people. The product didn't have to find its audience; the audience built the product spec by surfacing the same frustration repeatedly at meetups. That's Greg Eisenberg's framework in practice: build the audience, let it become a community, then build what the community needs. You launch into warm distribution instead of cold outreach.
In a world where anyone can build anything with AI tools, what actually determines success?
Marketing, distribution, customer acquisition, and revenue generation. When technical barriers are eliminated for everyone, engineering ability stops being a differentiator - it becomes table stakes. The ID345 community surfaces this clearly: the people who succeed with vibe coding aren't necessarily the strongest engineers. They're the ones who figure out how to get to a customer. Sam Altman's 'era of the ideas person' is real, but the idea alone isn't enough. The edge is the ability to get the idea in front of a paying customer before your window closes - and that's a distribution and marketing problem, not a technical one.
What's the bear case for vibe coding?
AI is genuinely coming for jobs, and that's real - not hype. Anyone who isn't actively learning and staying flexible about how they work is in a genuinely difficult position. The machines can do a lot of things. The bear case for vibe coding specifically is that the ease of starting creates a massive amount of unfinished projects and unrealistic expectations - people expect a deployable production application and get a clickable prototype. The gap between those two things is still real, and it's not going away just because the tools are faster. The 80-90% wall is a structural feature of the category, not a bug to be patched.