#Foreword — Why a Practitioner-to-App Field Manual Now

Carta's 2025 Solo Founders Report shows that solo-founded companies now account for over a third of all new US startups — the highest share in more than fifty years.^[1] AI-enabled founders are generating between $1.5 million and $10 million in revenue per employee, up from roughly $150,000 per head in 2022.^[1] Y Combinator's Winter 2026 batch was 88% AI-first across 199 companies, and 56 of them shipped fully autonomous agents that complete jobs without a human in the loop.^[2] When the cost of building a product collapses, the remaining bottleneck moves to distribution — and the founders most likely to own a vertical are the ones who have already lived inside the workflow.

This is the moment for the practitioner-to-app pipeline. A therapist who has spent four hours a night writing SOAP notes already knows the pain that Confidant Notes solves.^[3] A dentist who has tracked a month of unanswered after-hours phone calls already knows the gap that DentiVoice fills.^[4] These founders are not building from market research; they are building from a problem they have lived. As DentalBase founder Muhammad Abdel-Rahim wrote in April 2026: "I had the problem for six years before I built anything. Most founders research their customer. I was mine. There is a difference between knowing a problem exists and having lived inside it every single day for years. That difference is the whole product."^[4]

This manual is the operator's view of the practitioner-to-app pipeline as of May 2026. It covers the vibe-coding platform stack (Lovable, Bolt.new, v0, Cursor, Replit, and the recently absorbed Windsurf); the standard production architecture (Lovable → Supabase → Stripe → GitHub → Vercel) at $50^[5]-70 per month before revenue; six practitioner-founder examples spanning therapy, dentistry, and clinical documentation; the 90-day playbook from idea to first paying customers; five practitioner archetypes; and seven 2026-2030 predictions.^[5][6][7] Sources are vendor pricing pages, Anthropic and Stripe MCP documentation, founder essays from Medium, and indie-hacker playbooks from Awesome Agents, IndieRadar, and Foundra.^{[6][8][9][4][1][2]}

#Executive Summary — Top-Line Facts + Verdict

The vibe-coding category has produced more than $48 billion in combined valuation across four leaders.^[10] Lovable closed a $330M Series B in December 2025 at a $6.6B valuation led by CapitalG and Menlo Ventures — five months after a $200M Series A at $1.8B (a 3.7× step-up).^[11][10] Cursor (Anysphere) sits at $29.3B as of November 2025, with a SpaceX option deal that values an acquisition at $60B announced in April 2026.^[12][10] Replit raised a $400M Series D at $9B in March 2026 led by Georgian Partners.^[12][13] Vercel raised $300M at $9.3B in September 2025, driven significantly by v0's growth from 3.5M users in September 2025 to more than 6M developers by March 2026.^[11][13]

Revenue trajectories. Lovable hit $100M^[10][10] ARR in eight months, then doubled to $200M^[10][10] ARR four months later — what CEO Anton Osika calls "faster than OpenAI, Cursor, Wiz, and every other software company in history."^[10] Bolt.new went from $0 to $40M ARR in five months.^[11] Replit went from $2.8M at the start of 2025 to $150M ARR by September and is targeting $1B in run-rate revenue by end of 2026.^[13]

The standard production stack. Lovable → Supabase → Stripe → GitHub → Vercel runs $50-70 per month before revenue.^[5] Lovable Pro at $25/month plus Vercel Pro plus a custom domain covers the platform layer.^[6] Supabase free tier covers the first 50,000 monthly active users with a 500MB database.^[14] Stripe charges only on transactions (2.9% + $0.30) with no monthly fee.^[9] Real production cost on Supabase rises to about $125/month once you upgrade to Pro plus a larger compute instance.^[15] Claude Code costs average $13 per developer per active day or $150-250 per developer per month across enterprise deployments, with under $30 per active day for 90% of users.^[8]

Practitioner-founder cohort. Confidant Notes (built by therapist Jesse Saland LCSW-R, runs entirely on macOS, $399 one-time + optional maintenance fee) cuts therapist documentation time by 80%.^[3] DentalBase / DentiVoice (built by dentist Muhammad Abdel-Rahim) is an AI dental receptionist designed for the moment a patient calls during a procedure.^[4] Teja is AI-native practice management for therapists (Solo + Growing + Group tiers).^[7] Nora Health UK provides on-device GDPR-compliant clinical documentation for £24-£79/month.^[16] Omnira (built by Cyriac Ndo Zeh's Nidrosoft) is an AI-native dental practice OS with five autonomous AI agents on a Next.js 16 + React 19 + Supabase + Stripe stack.^[17]

The 90-day playbook. 50/50 build/marketing split. Two distribution channels (Twitter/X + SEO for most). Don't quit your job until $3^[1]-5K^[1] MRR. Bannerbear's Jon Yongfook hit $10K^[1][1] MRR in 12 months via 50/50 split plus integrations bringing 8-12 customers per launch.^[1] Senja.io's Wilson Wilson grew 0→20,000 Twitter followers, with 70% of Senja's early customers coming from Twitter.^[1] Median time from first revenue to $1M ARR across ChartMogul's 6,520-company dataset is 2 years 8 months — though AI-native startups sometimes do it in under 12 months.^[1]

Verdict. The bottleneck is no longer the code. The bottleneck is the combination of (a) a real practitioner who has lived inside a specific workflow long enough to know which 18% bookings improvement actually moves the buyer, and (b) a 90-day distribution motion that puts the product in front of peers. The practitioner-to-app pipeline is the structural arbitrage of 2026.

#Part I — The Vibe-Coding Stack: Lovable, Bolt, v0, Cursor, Replit

The vibe-coding category has settled into three distinct tiers as of 2026.^[18] First, AI-assisted IDEs (Cursor, Replit) — you're still writing code, but with an AI copilot that understands your full codebase. Second, AI app generators (Lovable, Bolt.new, v0) — you describe what you want and get generated code you can modify. Third, visual no-code platforms with AI (Bubble) — traditional no-code with AI features bolted on.

Lovable (Stockholm, founded 2024 by Anton Osika and Fabian Hedin) is the unicorn of the AI app generator tier. The company hit $100M^[11][11] ARR in eight months from launch, then doubled to $200M^[11][11] ARR four months later, and was estimated at $300M ARR by January 2026 and $400M ARR by February 2026.^[11][10][13] More than 25 million projects have been created on the platform with 100,000 new ones launched daily; the team grew from 45 to 146 employees in late 2025 with plans to reach 350 by end of year.^[12][19] The product generates full-stack React applications with Supabase backend integration; pricing is Pro $25^[19]/month (100 monthly credits + 5 daily up to 150/month + custom domains + remove badge) and Business $50^[6]/month (SSO + team workspace + design templates + role-based access + security center).^[6] Credit packages scale from 100 credits ($25/month) to 10,000 credits ($2,250/month) on Pro.^[20] Enterprise clients include Klarna, HubSpot, and Photoroom.^[19]

Bolt.new (StackBlitz, launched October 2024) is the WebContainers-powered browser-native app builder.^[19] WebContainers is a full Node.js environment that runs entirely inside the browser tab — Bolt can install npm packages, run a backend server, and execute real code without local setup.^[19] Bolt went from $0 to $4M ARR in its first 30 days, $20M in three months, and $40M^[19][19] ARR by March 2025; by mid-2025 it had crossed 5 million users and was adding over 1 million new users per month.^[13] StackBlitz raised a $105.5M Series B led by Emergence Capital and GV in January 2025 at approximately $700M valuation.^[11]

v0 by Vercel (founded 2023, 26 employees as of February 2026) sits in a different position than Lovable or Bolt.^[21] It generates React + shadcn/ui + Tailwind components and full-stack Next.js applications, but it's designed to slot into a developer's existing workflow.^[19] The January 2026 rebrand from v0.dev to v0.app tracked with significant product expansion: sandbox-based runtimes, Git panel integration for branches and PRs directly from chat, and database integrations with Snowflake and AWS.^[19] v0's code editor is now powered by VS Code on web and iOS, and v0 Max (the new default for all users) runs on Claude Opus 4.5 with 20% cheaper pricing.^[21] v0 reached 6 million developers by March 2026 with more than 80,000 active teams; over 50% of v0's revenue now comes from Teams and Enterprise plans.^[11][13]

Cursor (Anysphere) is the AI-assisted IDE leader at $29.3B valuation as of November 2025.^[10] The company employs 403 people in San Francisco and operates in 32 countries.^[22] Cursor pricing: Pro $20/month (500 fast requests + unlimited slow), Ultra $200/month (20× usage), Teams $40/user/month.^[23] Long-running agents launched at cursor.com/agents in February 2026 for Ultra/Teams/Enterprise plans.^[22] Composer 1.5 went into general availability with 3× usage on individual plans versus Composer 1.^[22] Opus 4.6 fast mode arrived on Cursor in February 2026 priced at $30 input / $150 output tokens.^[22] Per Cursor's corporate disclosure, in April 2026 SpaceX announced a deal that gives it the option to acquire Cursor (Anysphere)^[22] for $60B or pay $10B for collaboration.^[12]

Replit (founded 2016 by Amjad Masad) raised a $400M^[12][12] Series D^[12] at $9B^[12][12] in March 2026 led by Georgian Partners with Coatue, Andreessen Horowitz, Craft Ventures, Accenture Ventures, and angels Shaquille O'Neal and Jared Leto.^[12] Replit serves 35 million users in 200+ countries with 750,000+ businesses on the platform.^[13] Revenue went from $2.8M at the start of 2025 to $150M ARR by September, with Replit targeting $1B in run-rate revenue by end of 2026.^[13] Replit Agent 4 runs 10× faster than its predecessor and adds parallel-agent workflows.^[13] Pricing: Free + Replit Core $25/month (or $20/month annual, $25 monthly credits) + Replit Pro $100/month (or $95/month annual, $100 monthly credits) + Enterprise.^[24]

Windsurf (formerly Codeium) was acquired by Cognition in July 2025 after OpenAI's $3 billion deal to acquire the company fell through.^[12] Windsurf had Pro at $15/month with unlimited AI agent access.^[23] Emergent is a YC W24 fastest-growing platform founded by twin brothers Mukund and Madhav Jha; its January 2026 Series B^[23] raised $70M^[12][12] from Khosla Ventures and SoftBank Vision Fund 2 with participation from Prosus, Lightspeed, Together, and Y Combinator.^[12] Wix bought Base44 (a six-month-old solo-founder bootstrapped startup) for $80M in 2025.^[12]

#Part II — Production Stack Architecture

The standard practitioner-to-app production stack is Lovable → Supabase → Stripe → GitHub → Vercel, totaling $50-70 per month before revenue.^[5] Each component is the de facto standard for its layer, and each component is now MCP-server-accessible by AI agents from Claude Code, ChatGPT, Cursor, and the rest of the supported-client cohort.^[25][9]

Lovable is the front-end + LLM-orchestration layer. Lovable Pro at $25^[6]/month covers 100 monthly credits, custom domain support, the ability to remove the "Edit with Lovable" badge, and Code Mode for direct code editing.^[6] Credit packages scale linearly: 100 credits = $25/month, 200 = $50, 400 = $100, 800 = $200, 1,200 = $294, 2,000 = $480, 5,000 = $1,125, 10,000 = $2,250.^[20] Business plan (which adds SSO + team workspace + role-based access + security center) doubles the per-credit cost.^[20] Top-up credits are available in 50-credit increments at $15 per 50 (Pro) or $30 per 50 (Business), valid for 12 months.^[20] Unused credits roll over monthly.^[20]

Supabase is the database + auth + storage + realtime + edge functions layer. The Free plan provides 50,000 monthly active users, 500MB database size, 5GB egress, 1GB file storage, and 500K^[14] edge function invocations — though projects pause after 1 week of inactivity (with a 2-active-project limit).^[14] The Pro plan at $25/month provides 100,000 MAUs (then $0.00325 per additional MAU), 8GB database disk (then $0.125^[14]/GB), 250GB egress (then $0.09/GB), 100GB file storage, and 2 million edge function invocations.^[14] The Team plan at $599/month adds centralized billing, organization-level permissions, SOC 2 compliance, and 14-day automatic backups; Enterprise adds HIPAA add-on, AWS PrivateLink, and ISO 27001.^[14] Real production cost is approximately $125/month (Pro plus a larger compute instance), not the headline $25.^[15]

Row Level Security (RLS) is Supabase's defining security primitive.^[26][27] RLS is enabled by default on tables created via the Table Editor; tables created via raw SQL must have RLS enabled manually with ALTER TABLE ... ENABLE ROW LEVEL SECURITY.^[26] Supabase provides helpers including auth.uid() (which returns the user UUID, or null for unauthenticated requests) and auth.jwt() (which returns the full JWT including raw_app_meta_data).^[26] Service keys bypass RLS — they should never be used in the browser or exposed to customers, but they are useful for administrative tasks.^[26] In Postgres 15 and above, views can be made to obey RLS policies of the underlying tables when invoked by anon and authenticated roles by setting security_invoker = true.^[26] Without RLS enabled and policies defined, no data is accessible via the API when using a publishable key.^[26]

Stripe is the billing + subscription + tokenization layer. Stripe Agent Toolkit (Python and TypeScript) lets AI agents interact with Stripe APIs through function calling, with examples for OpenAI Agent SDK, LangChain, CrewAI, and Vercel AI SDK.^[28] Stripe MCP server runs at https://mcp.stripe.com with OAuth, exposing tools for get_stripe_account_info, create_customer, create_invoice, create_payment_link, create_subscription, create_refund, search_stripe_resources, and more.^[9] Local MCP via npx -y @stripe/mcp --api-key=YOUR_STRIPE_SECRET_KEY is supported.^[9] Restricted API Keys (rk_*) are recommended for granular agent permissions; tool availability is determined by RAK permissions.^[9][28] Stripe processing is 2.9% + $0.30 per transaction with no monthly fee.^[7] The @stripe/agent-toolkit/ai-sdk package integrates Stripe's billing infrastructure directly with Vercel's AI SDK.^[29][28]

Vercel is the deployment + serverless functions + edge layer. Vercel MCP server runs at https://mcp.vercel.com with OAuth and supports 12 reviewed/approved AI clients: Claude Code, Claude.ai/Claude for Desktop, ChatGPT, Codex CLI, Cursor, VS Code with Copilot, Devin, Raycast, Goose, Windsurf, Gemini Code Assist, and Gemini CLI.^[25] Setup via npx add-mcp https://mcp.vercel.com auto-detects installed AI clients and configures Vercel MCP for each one.^[25] Tools include searching documentation, managing projects/deployments, and analyzing deployment logs.^[25]

Anthropic Claude pricing sits at the bottom of the cost stack for the AI inference itself.^[30] Claude Opus 4.7 (and 4.6) cost $5/MTok input and $25/MTok output, with $0.50/MTok cache hits and $6.25-$10/MTok cache writes.^[30] Sonnet 4.6 is $3/MTok input and $15/MTok output.^[30] Haiku 4.5 is $1.25/MTok input and $5/MTok output.^[30] Claude Code costs across enterprise deployments average $13 per developer per active day or $150-250 per developer per month, with <$30 per active day for 90% of users.^[8] The /usage command provides per-session token statistics; MAX_THINKING_TOKENS=8000 reduces extended thinking budget; CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 enables agent teams (disabled by default).^[8]

#Part III — The Domain-Expert Founder Cohort

Six practitioner-founder examples illustrate the shape of the 2026 vertical-AI cohort.

Confidant Notes is the canonical "I built the tool I wished existed" archetype. CEO and co-founder Jesse Saland LCSW-R is a practicing therapist; Confidant runs entirely offline on macOS 13+ Apple Silicon devices.^[3] Audio is captured, transcribed, and processed entirely on the device with no cloud, no servers, and no AI training on client sessions.^[3] Because Confidant has zero access to client data, no Business Associate Agreement (BAA) is needed; PHI never touches a server.^[3] Therapists using Confidant cut documentation time by up to 80%.^[3] Pricing is one-time $399 plus optional maintenance fee for access to new AI models as they're released.^[3] As Saland writes: "Most software protects your data with promises. Confidant protects it with architecture."^[3]

DentalBase / DentiVoice is the canonical "founder still practicing" archetype. Co-founder Muhammad Abdel-Rahim — a practicing dentist — tracked, for one full month in 2022, how many new patients his practice was losing to unanswered calls.^[4] Every existing solution (better scheduling apps, patient portals, callback systems) addressed the problem from outside the practice rather than from inside it.^[4] Abdel-Rahim built DentiVoice to be ambient — to work without any staff decision to activate it, to simply be there every time the phone rings.^[4] The first version had too many decision branches and too many clarifying questions; it was stripped back to two jobs (answer questions, book appointments) with everything else routed to a human with a flag.^[4] As Abdel-Rahim writes: "The niche is not a constraint. It is the reason a generalist cannot walk in and take the market. You need to have stood in the practice to understand what the practice actually needs."^[4] He still practices dentistry to keep the product honest.^[4]

Teja is the canonical "AI-native vertical SaaS" archetype. The product is AI-native practice management for mental health professionals — built from the ground up with AI at its core rather than bolted on.^[7] Teja claims 60% less time on admin work; AI Progress Notes save 5-10 hours per week and turn session recordings into structured clinical notes in under 2 minutes (SOAP, DAP, BIRP, GIRP, PIRP formats).^[7] Golden Thread Treatment Planning creates SMART goals + tracks progress visually + auto-generates insurance compliance reports, reducing claim denials.^[7] Teja's pricing model is meaningful: connect your own Stripe account; Teja charges $0 processing fees; you pay only Stripe's standard 2.9% + $0.30.^[7] Modality-aware AI for CBT, DBT, EMDR, and more.^[7]

Nora Health UK is the canonical "GDPR-by-architecture" archetype. The platform is purpose-built for UK private practitioners with on-device session note generation; audio capture, transcription, and note generation all live entirely on the practitioner's hardware with no cloud inference and no external APIs.^[16] Nora is registered with the UK Information Commissioner's Office and built around UK data protection law.^[16] Pricing: Free plan with up to 5 active clients + Pro £24/month (full clinical intelligence) + Pro+ £79/month (5 practitioners + practice dashboard + team management).^[16] Specialisation profiles for trauma, adolescent psychology, couples therapy, CBT, and more.^[16]

Omnira is the canonical "team-around-domain-expert" archetype, illustrated through Cyriac Ndo Zeh's Nidrosoft engineering team building a dental practice operating system with five autonomous AI agents (Stella, Vera, Relay, Aria, Otto) for scheduling, claims processing, patient communication, clinical documentation, and operations.^[17] The technical stack is illustrative: Next.js 16 with App Router and Turbopack + React 19 + TypeScript 5 + Tailwind CSS 4 alongside CSS Modules + Framer Motion 12 for animations + Zustand 5 for state + Supabase auth/database + Sonner toast notifications + Twilio SMS + Stedi insurance + Stripe payments + Retell.ai voice.^[17] The custom design system "Omnira UI" includes 34 UI primitives and is dark-mode-first.^[17] Settings has 13 configuration sections plus a 7-step onboarding wizard.^[17] Omnira's thesis: replace 5-10 disconnected tools (Dentrix scheduling, Open Dental charting, Weave communication, Dental Intelligence analytics) with one unified platform.^[17]

The unfair advantage across all six is being the customer first. As Abdel-Rahim writes: "I know what a dental practice owner reads. I know which language lands and which sounds like vendor noise. I know that 'improved ROI' moves no one, but 'new patient bookings up 18 percent in month one' moves everyone."^[4] This advantage only holds if the founder stays current with the problem; founders who were the customer six years ago and have since moved on lose the edge fast.^[4]

#Part IV — The 90-Day Playbook

The 90-day playbook for going from idea to first paying customers — distilled from indie hackers, YC W26 distribution lessons, and the Lovable founder cohort — is structured in five 14-15-day phases.^[2][1][31]

Week 0: Validate the problem. Identify 20 people who have the specific problem you want to solve.^[1] If you cannot find 20, pick a different problem. The golden rule from indie hackers: "If they aren't ready to pay now, put down a deposit, or hand over their credit card—we don't build it."^[32] Send 20 cold emails or social DMs with a simple question: "Are you struggling with X? I'm building a solution and will give you 6 months free in exchange for a 15-minute call."^[32] If you get 3-5 "yes" replies, build the landing page. If you don't, the wedge isn't real yet.^[32]

Days 1-14: Foundation. Define your ICP (ideal customer profile). Build your landing page (Carrd, Webflow, or Framer is fine — perfection is not the goal). Set up analytics (Plausible or PostHog at $0^[1]-20/month). Set your pricing. The pricing rule: charge from day one. Free users give you vanity metrics; paying users give you signal.^[1][31]

Days 15-30: Pre-launch. Start building in public on Twitter/X (3-5 posts per week, 95% value, 5% product mentions).^[1] Wilson Wilson of Senja.io grew 0→20,000 followers by sharing daily progress updates; 70% of Senja's early customers came from Twitter.^[1] Join 3-5 relevant communities (subreddits, Discord servers, Slack groups) — the Reddit 90/10 rule applies: 90%^[1] value, 10% product, with weeks of genuine participation before any self-promotion.^[1] Invite 10-20 people to a private beta and collect testimonials; ask for early wins and user quotes.^[31] Pre-sell a "founding plan" at 50% lifetime discount; goal is 5 payments in 7 days.^[33] If money hits the bank, build the MVP using AI tools and a boilerplate (Lovable + Supabase + Stripe + GitHub + Vercel).^[5] If nobody pays, refund what you got, kill the idea, and move to the next one.^[32]

Days 31-45: Launch. Execute the launch plan: Product Hunt + push hard on all channels for 2 weeks.^[31] Line up 5-10 people who will share your launch.^[31] Track MRR, churn, and CAC; ignore everything else until you hit $1K MRR.^[1]

Days 46-75: Iterate. Analyze what worked and what didn't.^[31] Double down on your best-performing channel.^[31] Start your content / SEO strategy. The Zapier programmatic SEO model is the high-bar version: 50,000+ integration landing pages producing 16M monthly organic visitors.^[1] DelightChat grew from 600 to 240,000 search impressions per month in 90 days by creating 300+ pages targeting "best Shopify apps for [category]" using a CMS API.^[1] You don't need that scale, but the principle applies: find a keyword pattern, templatize the page, and produce dozens of variations.^[1]

Days 76-90: Systematize. Build repeatable processes for your best channels.^[31] Automate what you can; delegate what you cannot.^[31] $500/month paid acquisition allocates 90% to Google Ads ($450) and 10% to experimental channels ($50 ^[31]— Reddit ads, X ads, newsletter sponsorships, podcast ads); kill what doesn't work within 2 weeks.^[31]

The 50/50 rule. From the Awesome Agents Solo SaaS Founder playbook: split your time 50/50 between building and marketing.^[1] This ratio is non-negotiable. The 92% SaaS startup mortality rate is overwhelmingly driven by distribution failures, not product failures.^[1] Bannerbear's Jon Yongfook hit $10K MRR in 12 months via 50/50 split + integrations bringing 8-12 customers per launch; he maintained excellent developer documentation and published tutorial content that ranked well in search.^[1]

The two-channel commitment. Pick two distribution channels and commit for 90 days.^[1] For most founders: Twitter/X plus SEO content.^[1] Reddit converts higher than LinkedIn or Facebook for B2B SaaS but only if you've spent weeks participating genuinely before any self-promotion.^[1]

The exit-day-job threshold. Don't quit your job until $3-5K MRR.^[1] That's the range where the business can sustain you through a bad month.^[1] ChartMogul's analysis of 6,520 companies puts the median time from first revenue to $1M^[1][1] ARR at 2 years and 8 months; AI-native startups sometimes do it in under 12 months — ShiftNex (Lovable cohort) reached $1M ARR in five months with 5,000 healthcare users.^[1][5]

Distribution-first. Before writing a single line of code, write the distribution plan first.^[34] If you don't know where the audience already lives, you don't have a wedge. Outputs do the marketing — every artifact your tool creates should sell the tool for you.^[34] Plug into existing platforms: Slack bot + Notion integration + Google Sheets add-on + Figma plugin + Chrome extension built in a weekend with Cursor or Claude Code.^[34]

#Part V — Funding Velocity + Acquisition Cadence

The vibe-coding category has produced more than $48 billion in combined valuation across four leaders (Lovable + Cursor + Replit + Vercel).^[10] The funding velocity has been the fastest in software history.

Lovable raised more than $500M in 2025 across two rounds.^[10] Series A was $200M led by Accel in July 2025 at $1.8B valuation; participants included Creandum, Klarna founder Sebastian Siemiatkowski, ElevenLabs founder Mati Staniszewski, and Synthesia founder Victor Riparbelli.^[10] Series B was $330M led by CapitalG (Alphabet/Google) and Menlo Ventures' Anthology fund in December 2025 at $6.6B valuation — a 3.7× step-up in five months.^[11][10] Other Series B investors: NVIDIA NVentures, Salesforce Ventures, Databricks Ventures, HubSpot Ventures, Atlassian Ventures, Deutsche Telekom T.Capital, Khosla Ventures, DST Global, EQT Growth, and Kinship Ventures (backed by actress Gwyneth Paltrow).^[10]

Cursor (Anysphere) is at $29.3B valuation as of November 2025 with $2.3B Series B raised.^[10] Per Cursor's corporate disclosure, in April 2026 SpaceX struck a deal that gives it the option to acquire Cursor (Anysphere)^[22] for $60B or pay $10B for collaboration.^[12] The collaboration strengthens SpaceX's position in the AI coding race against top labs building advanced coding tools.^[12]

Replit raised $400M^[12][12] Series D^[12] in March 2026 at $9B^[12][12] valuation led by previous investor Georgian Partners with Coatue, Andreessen Horowitz, Craft Ventures, Accenture Ventures, and angel investors Shaquille O'Neal and Jared Leto.^[12]

Vercel raised $300M^[11][11] at $9.3B^[11][11] valuation in September 2025; the round was significantly driven by v0's growth, with Vercel's overall revenue growing roughly 82% year-over-year ahead of the Series F.^[11]

Acquisition cadence. Cognition acquired Windsurf in July 2025 after OpenAI's $3 billion deal to acquire the company fell through.^[12] One month before the Cognition deal, Wix bought Base44 — a six-month-old solo-founder bootstrapped startup — for $80 million.^[12]

Smaller-tier funding. StackBlitz/Bolt raised $105.5M^[11][11] Series B^[11] led by Emergence Capital and GV in January 2025 at approximately $700M valuation.^[11] Emergent (YC W24, founded by twin brothers Mukund and Madhav Jha) raised $70M^[11][11] Series B^[11] in January 2026 from Khosla Ventures and SoftBank Vision Fund 2 with Prosus, Lightspeed, Together, and Y Combinator participating.^[12]

TAM projections. AgentMarketCap estimates the vibe-coding market at $4.7B in 2026 on a trajectory to $12.3B by 2027 (38% CAGR).^[11] Other estimates project $325B by 2040 (36.79% CAGR from $2.96B in 2025).^[13] These numbers likely undercount the actual market impact — they measure tool revenue, not the value of software created.^[11]

#Part VI — Five Practitioner Archetypes

Five archetypes describe how practitioners-turned-founders connect domain expertise to capital and outcomes in 2026.

Archetype 1 — The Still-Practicing Founder. DentalBase (Muhammad Abdel-Rahim, practicing dentist), Confidant Notes (Jesse Saland LCSW-R, practicing therapist).^[3][4] The thesis: keep practicing as the moat; the product is validated against real-time pain. The risk is bandwidth — the founder splits time between practice and product, which limits scale.^[4]

Archetype 2 — The Full-Time Pivoter. Teja (mental-health-professional team that left clinical practice to build full-time), ShiftNex (Lovable cohort, healthcare).^[7][5] The thesis: exit the practice once early product-market-fit signal hits ($1K^[5][5]-$3K^[5][5] MRR). The risk is losing the customer-empathy advantage; mitigation requires staying close to the practitioner community via design partners and customer councils.^[4]

Archetype 3 — The Team-Around-Domain-Expert. Omnira (Nidrosoft engineering team partnered with practitioner-founder Cyriac Ndo Zeh).^[17] The thesis: a non-practitioner builder partners with a practicing founder for domain validation; the team-around model captures the technical depth needed for AI-agent architectures while preserving the practitioner's authority. The risk is alignment — the technical lead must defer to the domain expert on product calls that contradict engineering preference.^[17]

Archetype 4 — The Acqui-Target. Base44 (six-month-old solo founder → $80M Wix acquisition), Junglytics (acquired by Carbon6 August 2024).^[12] The thesis: build a wedge product over 6-12 months and sell to an incumbent (Wix, RealPage, Entrata, Helium 10's Assembly parent). The risk is over-fitting to a single acquirer's product roadmap; mitigation requires a parallel direct-sales motion.

Archetype 5 — The Indie Hacker. Bannerbear (Jon Yongfook, $10K^[1][1] MRR in 12 months), Senja (Wilson Wilson, 70%^[1] of customers from Twitter), QuickTables ($100K+ annual revenue from a 60-day quit-the-job constraint).^[1][5] The thesis: solo or duo team, 50/50 build/marketing split, build-in-public on Twitter/X, no outside capital. The risk is the founder's inability to scale beyond what one person can ship; mitigation requires accepting a $5-50K MRR ceiling or hiring once revenue justifies it.^[33]

The investor cohort that recurs across multiple winners includes Accel, Khosla Ventures, Andreessen Horowitz, CapitalG (Alphabet), Menlo Ventures, Coatue, Sapphire Ventures, Greylock, Y Combinator, and Founders Fund.^[10][12] The distribution channels that recur include Twitter/X (Senja, Bannerbear, Lovable), Product Hunt (most launches), Indie Hackers community, niche subreddits, and Reddit 90/10 participation.^[1][31]

#Part VII — Seven Predictions, 2026-2030

Prediction 1. Practitioner-built apps cross $1B^[7][7] aggregate ARR by 2027. With 5.6M^[7] solopreneurs earning $100K^[7][7]+ in 2025 plus accelerating vibe-coding tooling, the conditions are in place for one or more practitioner-founder verticals (therapy, dentistry, accounting, GC) to produce a $50M+ ARR vendor by 2027.^[7][5]

Prediction 2. 5+ vertical AI vendors raise $50M^[10][10]+ rounds from practitioner-founder cohort by 2027. Following the Carry → Lettuce → Confidant → Teja path, expect VC institutional capital flowing specifically into vendors with practitioner founders rather than ex-FAANG-engineer founders.^[10]

Prediction 3. Apple/macOS local-AI vertical wins outpace cloud-AI in HIPAA/GDPR-sensitive categories. Confidant Notes (macOS-only) and Nora Health (on-device GDPR) are the leading edge; expect 3+ new entrants to copy this architecture in mental health + dental + legal + financial planning verticals by 2027.^[3][16]

Prediction 4. Lovable IPOs at $20B^[13][13]+ valuation by 2028. With $400M^[13][13] ARR by February 2026 and continued doubling cadence, $1B^[13][13]+ ARR by 2027 is plausible; an S-1 at 20× ARR puts the company in the $20B-$30B range.^[13][10]

Prediction 5. Practitioner-to-app accelerator/incubator emerges. The vertical-specific YC equivalent — possibly run by a vertical-platform vendor (Lovable, Replit) or a healthcare/legal-specific firm — will fund 50-100 practitioner founders per year by 2028.^[5]

Prediction 6. 50%^[4] of new vertical SaaS in regulated industries built by practitioners-turned-founders by 2028. The combination of vibe-coding economics + practitioner customer-empathy moat + AI-CPA / AI-attorney compliance tooling will flip the founder demographic for regulated-vertical SaaS.^[4][7]

Prediction 7. Existing EHRs/PMS acqui-hire practitioner founders for 50%^[12]+ of vertical AI bolt-ons by 2028. Following the Entrata → Colleen and RealPage → Knock pattern, incumbents will systematically acquire practitioner-founder startups to capture both the product and the credibility-with-peers distribution motion.^[12]

#Glossary + Related Research

Glossary

• ARR — Annual Recurring Revenue.

• BAA — Business Associate Agreement (HIPAA).^[3]

• EHR — Electronic Health Record system.^[7]

• GDPR — General Data Protection Regulation (EU/UK).^[16]

• ICP — Ideal Customer Profile.^[31]

• MAU — Monthly Active User.^[14]

• MCP — Model Context Protocol (Anthropic, open standard).^[25][9]

• MRR — Monthly Recurring Revenue.^[1]

• OAuth — Open Authorization protocol.^[25][9]

• PMS — Practice Management Software.^[17]

• RLS — Row Level Security (Postgres).^[26][27]

• RAK — Restricted API Key (Stripe).^[28]

• SaaS — Software as a Service.

• SOAP — Subjective / Objective / Assessment / Plan note format.^[7]

• TCO — Total Cost of Ownership.^[18]

• Vibe coding — Andrej Karpathy's February 2025 coinage; software development by handing high-level intent to AI; Collins Dictionary Word of the Year 2025.^[11]

• WebContainers — StackBlitz technology that runs full Node.js inside a browser tab.^[19]

Related Research

This paper does not cover four threads worth their own treatment: (1) the Lovable→Supabase→Stripe Production Stack Audit — security review, RLS policy patterns, common Lovable-deployed app vulnerabilities (the March 2025 Replit-discovered Lovable-built Supabase access-control issues are the canonical case); (2) the Practitioner-Founder Mental Health Cohort 2027 — therapy-vertical specifically, with EHR migration patterns from SimplePractice/TherapyNotes/Headway to Confidant/Teja; (3) the Indie Hacker $10K MRR Distribution Playbook — Bannerbear-style integration-led growth with a step-by-step Slack/Notion/Sheets/Figma/Chrome integration ship roadmap; (4) the Vertical-AI Acqui-Roll-Up Predictability Study — pricing benchmarks for $3-50M ARR vertical AI exits across Wix→Base44, Cognition→Windsurf, RealPage→Knock, Entrata→Colleen, Carbon6→Junglytics.

#References