# Okane Land

> Real numbers for solo AI builders. An independent publication and community for vibe coders, founders, and marketers making money with AI. Curator voice: we feature and vet other people's AI builds, test the tools, and do the cost and pricing math nobody else does. Show your work, receipts required.

Published by Okane Land. The site is organized into seven sections: an AI-news column verified to primary sources and published when a story earns it (The Signal), builder interviews (The Studio), vendor-neutral tool how-tos (The Primer), a curated tool shortlist (The Palette), hands-on tested reviews (The Proof), researched cost and comparison analysis (The Study), and the money room of income, pricing, selling, and growth (The Ledger). This file follows the llms.txt convention so AI agents can find the pages worth reading.

## Start here
- [Start here](https://okaneland.com/start-here/): what Okane Land is, who it serves, and how the sections fit together.

## The Signal: the AI-news column, verified to primary sources
- [Your users can pay your AI bill with ChatGPT. OpenAI keeps the meter.](https://okaneland.com/signal/openai-keeps-the-meter/): ChatGPT Plus and Pro users can now spend their plan in participating apps. What that does to a small AI product's costs, and what OpenAI still controls.
- [The frontier now ships twice. The second copy is not for sale.](https://okaneland.com/signal/the-frontier-twice/): Anthropic, Google and OpenAI each shipped a public model and a vetted-access copy in one week. What the gate asks for, and what it means for a solo builder.
- [A worm is after your AI keys. The install is the door.](https://okaneland.com/signal/the-install-is-the-door/): A self-spreading npm worm hit 400+ packages and steals the API keys in your AI tools, hiding in .claude and .vscode config. The install ran its code.
- [Nothing leaked. That is the part worth worrying about.](https://okaneland.com/signal/nothing-leaked/): Shared Claude chats hit Google. Nothing was breached: users published them. Share quietly became publish, and builders keep shipping the same gap.
- [An AI built a forum for AI agents. Permanence costs $1.](https://okaneland.com/signal/priced-permanence/): An agent handed a domain built a forum addressed to software. Posting is free; the one priced good is permanence. What that costs, and who actually paid.
- [The vibe coding split is a pricing event](https://okaneland.com/signal/vibe-coding-split-pricing-event/): The search data split vibe coding in two. The casual mode keeps the name; the disciplined mode got its own: agentic engineering. Two words end one price.
- [AI made code free. Accountability is the scarce part now.](https://okaneland.com/signal/ai-made-code-free/): Godot banned autonomous AI agents and said why: AI cannot take responsibility. The open-source AI bans are a repricing, and solo builders are on both sides.

## The Ledger: the money (income, pricing, selling, growth)
- [Stripe or a merchant of record: the checkout decision](https://okaneland.com/ledger/stripe-or-merchant-of-record/): Stripe vs Paddle, Lemon Squeezy, Polar, and Gumroad on a $19/mo product: the fee math six ways, plus the sales-tax liability Stripe leaves you holding.
- [How to price a B2B SaaS](https://okaneland.com/ledger/how-to-price-a-b2b-saas/): How to price a B2B SaaS: the three pricing models (seat, usage, value), cost-plus vs value-based, and why underpricing is the common failure.
- [How to get your first 10 users](https://okaneland.com/ledger/first-10-users/): Where builders actually got their first 10 users and customers: the receipts, the catch each story carries, and the plays that work with no audience.
- [Getting paid for AI work: deposits, deadlines, and the clients to refuse](https://okaneland.com/ledger/getting-paid-for-ai-work/): How AI freelancers get paid: the nonpayment base rate, deposit and milestone structure, new 30-day payment laws, and the client red flags that say walk.
- [Productize the service: turn bespoke AI work into one thing you can sell twice](https://okaneland.com/ledger/productize-your-ai-service/): How to turn bespoke AI work into one fixed-scope, fixed-price offer that repeats: Designjoy and Conversion Factory receipts, scope rules, and the token bill.
- [The demo is the cheapest part: scoping AI work so it does not lose money](https://okaneland.com/ledger/scoping-ai-projects/): Why AI client projects lose money in delivery rather than pricing: the demo trap, the hidden scope range, what scope creep costs, and the AI maintenance tax.
- [The retainer ladder: from one-off AI builds to income that repeats](https://okaneland.com/ledger/retainers-for-ai-work/): What AI and automation retainers really pay, the MSP pricing norms behind them, the maintenance math, and the tactics that survive the month nothing breaks.
- [The economics of a one-person AI business: what the MRR screenshots leave out](https://okaneland.com/ledger/economics-of-a-one-person-ai-business/): The research on what a one-person AI business actually keeps: gross margins, churn, card fees, and the gap between MRR and income. Evidence over hype.
- [Pricing your AI app: the psychology that holds up, and the tricks that fall apart](https://okaneland.com/ledger/pricing-psychology-that-holds-up/): The pricing psychology with real evidence behind it, the famous tactics that collapse on replication, and how to price your own AI product.
- [What to charge for AI work without lowballing yourself](https://okaneland.com/ledger/pricing-ai-work/): How to price AI builds, automations, and retainers so being fast does not make you cheap, and how to set a number you can defend.
- [How to land your first AI client](https://okaneland.com/ledger/first-ai-client/): Land your first AI client: what the small-business AI data says, where the work comes from, and how to turn a warm lead into a paid build.

## The Study: researched cost and comparison analysis
- [OpenAI gives six months. Anthropic gives two. Google gives weeks.](https://okaneland.com/study/ai-model-shutdown-ledger/): We logged 71 shutdown notices from OpenAI, Anthropic and Google. Median warning: 182, 63 and 28 days. What it means for the model you build on.
- [What it costs to run a vibe-coded app](https://okaneland.com/study/what-it-costs-to-run-an-app/): Vercel + Supabase versus a Hetzner or DigitalOcean VPS across three stages, priced from each vendor's own pages, with the fine print that flips the math.
- [The MCP tax: what each server costs before your first prompt](https://okaneland.com/study/the-mcp-context-tax/): We measured the tool-definition token cost of ten popular MCP servers, and when that cost hits in current Claude Code. Per-server table, keep-or-cut verdicts.
- [Blast radius: what your AI agent can destroy](https://okaneland.com/study/what-your-ai-agent-can-destroy/): A verified ledger of AI-agent security failures: runaway spend, a nine-second production wipe, MCP data leaks, malicious skills, and the guardrail for each.
- [The vibe-coding wins are real. So is the denominator](https://okaneland.com/study/after-the-vibecode-wins-and-denominator/): The vibe coding success stories are real. What the ARR figures hide, the acquisition fine print, and the denominator from Pieter Levels' own ledger.
- [We pinged 522 vibe-coded launches. Here is how many answered](https://okaneland.com/study/how-many-vibe-coded-launches-are-alive/): We HTTP-checked 522 vibe-coded apps from the Lovable and Bolt showcases. At least 1 in 9 is already dark. The dead floor, the age curve, full dataset.
- [The AI coding agent field guide (2026)](https://okaneland.com/study/ai-coding-agent-field-guide/): A survey of the 2026 AI coding agents: Claude Code, Codex, Antigravity, Cursor, Zed, OpenCode, and the Windsurf story. What each is for, and how to choose.
- [Is your vibe-coded app secure? Run this audit](https://okaneland.com/study/is-your-vibe-coded-app-secure/): What the security research actually says about AI-generated code, why vibe-coded apps leak, and an eight-point self-audit you run on your own app.
- [How to cut your AI coding bill](https://okaneland.com/study/cut-your-ai-coding-bill/): What AI coding actually costs in 2026, the subscription-vs-API break-even, and four vendor-documented ways to cut the bill. All prices dated.
- [AI agents do not choose what to sell. They fill in a number a human left blank.](https://okaneland.com/study/what-agent-economies-charge-for/): Seven agent economies read to primary sources: who set the price, who paid, and what the money bought. The agent supplies the number, a human supplies the good.
- [Is distribution really the moat? What actually defends an AI product](https://okaneland.com/study/distribution-moat-ai/): Is distribution the moat in AI? Gemini surged on bundling while Copilot stalled. Which kinds of distribution defend a product, and which wash away.
- [The $5,000 tell: AI vendors are selling you human QA for their own AI](https://okaneland.com/study/the-5000-dollar-tell/): An AI ad agent spent 173 credits ($57) on one 40-second ad. Above it: a $5,000/mo human-QA tier. What a vendor's own pricing says about AI quality.
- [The year Claude subscribers sued over limits is the year the limits mostly went up](https://okaneland.com/study/claude-code-limits/): Claude Code limits, mapped from 2,122 subscriber posts: the 5-hour window, the weekly caps, the published per-plan bands, the lawsuit, and the checklist.
- [Is vibe coding dead, or just renamed? What the record shows](https://okaneland.com/study/is-vibe-coding-dead/): Vibe coding isn't dead: adoption is at record highs. What's fading is the word, split into casual vibe coding and agentic engineering. Naming timeline, sourced.
- [The new SDLC with vibe coding: what Google's paper actually says](https://okaneland.com/study/new-sdlc-with-vibe-coding/): Google's New SDLC With Vibe Coding whitepaper, explained and fact-checked: the vibe coding to agentic engineering spectrum, and which of its numbers hold up.
- [What a customer actually costs: CAC by channel for a one-person AI product](https://okaneland.com/study/cac-by-channel-ai-product/): Real CAC by channel for a solo AI builder: paid search near $$800, SEO about $$205, owned channels near zero, the 3:1 payback bar, and the AI churn tax.
- [Perplexity Pro limits: the shrinking $20 plan, in 1,024 posts](https://okaneland.com/study/perplexity-pro-limits/): Perplexity Pro limits, mapped from 1,024 coded subscriber posts: three quota cuts in eight months, why people cancel, what Max adds, and what to verify first.
- [Does building in public actually pay?](https://okaneland.com/study/build-in-public-revenue/): The receipts on building in public: a zero-ad $80M exit, an inflated ARR confession, 0.06% follower conversion, a $500 median, and the gates that decide.
- [What a ChatGPT citation is worth, and which answer engine a small site can win](https://okaneland.com/study/what-a-chatgpt-citation-is-worth/): What an AI citation is worth, fact-checked: the conversion premium, why AI traffic is really ChatGPT, the 11% engine overlap, and why a small site can win.
- [Is vibe coding worth it? What the numbers say](https://okaneland.com/study/is-vibe-coding-worth-it/): Yes, with receipts: an $80M exit, measured speed gains, 2026 loyalty data, and the maintenance and security habits that separate winners. The numbers, cited.
- [AI takes two-thirds of venture money, and your odds are still one in six](https://okaneland.com/study/ai-startup-raise-math/): Should you raise VC for your AI product? 65% of venture money went to AI, ~20% dilution per round, a $3M ARR Series A bar, and one-in-six graduation odds.
- [The AI UGC ad math: $2 a video, and the bill nobody prices in](https://okaneland.com/study/ai-ugc-ad-economics/): What AI UGC ads really cost ($1 to $11 a video) vs human creators, whether they convert (an independent Ipsos study says no), and the rules vendors omit.
- [AI saves about 3% of your hours, and almost none of it reaches the money](https://okaneland.com/study/ai-productivity-roi-at-work/): The real ROI of AI for knowledge work: the task-level gains, the jagged frontier, the 2.8% real-world time saving, and why so little reaches the paycheck.
- [AI agents finish a third of the job, and the math says why](https://okaneland.com/study/do-ai-agents-work-yet/): Do AI agents work yet? What TheAgentCompany, CRMArena-Pro, tau-bench, and METR show, and the compounding-reliability math behind the autonomy gap.
- [Wrap an LLM, charge $20, lose money: the pricing trap the model owners hit first](https://okaneland.com/study/unit-economics-of-wrapping-an-llm/): Why flat-rate pricing on an LLM product loses money: the heavy-user tail, the flat-rate bias, and the usage-based pricing the big AI products switched to.
- [AEO and GEO: one real study, a pile of mythology, and a traffic cliff](https://okaneland.com/study/aeo-and-geo-what-the-research-says/): GEO and AEO, fact-checked against the research: the one controlled study, what moves AI citations, the llms.txt and schema myths, and the real traffic data.
- [AI coding: faster MVP, slower review, and the security bill nobody mentions](https://okaneland.com/study/does-ai-coding-make-you-faster/): AI coding speeds up a new build and taxes review, maintenance, and security. The research on where it helps, where it bills you, and what to do.

## The Primer: vendor-neutral tool how-tos
- [How to use OpenCode: the open-source terminal agent](https://okaneland.com/primer/how-to-use-opencode/): A practical guide to OpenCode, the open-source terminal AI coding agent: 75+ model providers or a local model, the cost, and how it compares to Claude Code.
- [How to use Replit Agent: from prompt to deployed app](https://okaneland.com/primer/how-to-use-replit-agent/): A practical guide to Replit Agent: the build-and-deploy workflow, 2026 pricing and effort-based checkpoint billing, and how it compares to Bolt and Lovable.
- [How to use Cline: the open-source agent in your editor](https://okaneland.com/primer/how-to-use-cline/): A practical guide to Cline, the open-source AI coding agent for VS Code, JetBrains, and the terminal: Plan and Act, checkpoints, local models, costs, and MCP.
- [How to use v0: generate the UI, get real code](https://okaneland.com/primer/how-to-use-v0/): A practical guide to v0, Vercel's generative-UI builder: what it builds, how to get the code out, the 2026 credit pricing, and how it compares to Bolt.
- [How to use Bolt.new: build an app from a prompt](https://okaneland.com/primer/how-to-use-bolt-new/): A practical guide to Bolt.new, StackBlitz's in-browser AI app builder: the token model and free tier, owning your code, and how it compares to Lovable.
- [How to use Zed: the fast, open-source editor with AI](https://okaneland.com/primer/how-to-use-zed/): A practical guide to Zed, the fast open-source editor: the AI agent panel, a local model via Ollama, external agents like Claude and Codex over ACP, and Zeta.
- [How to use LM Studio: run local LLMs with a GUI](https://okaneland.com/primer/how-to-use-lm-studio/): A practical guide to LM Studio: download a model in-app, run the local OpenAI- and Anthropic-compatible server, and connect VS Code, Codex, and Claude Code.
- [How to use Aider: AI pair programming in your terminal](https://okaneland.com/primer/how-to-use-aider/): A practical guide to Aider, the terminal AI pair programmer: install, run it in a git repo, the in-chat commands, architect mode, and local models via Ollama.
- [How to use Ollama: run AI models on your own machine](https://okaneland.com/primer/how-to-use-ollama/): A practical guide to Ollama: install it, run a model, pick one that fits your RAM, wire it into your editor and agents via the local API, and Ollama Cloud.
- [How to make your product citable: an AEO walkthrough](https://okaneland.com/primer/make-your-product-citable-aeo-walkthrough/): A hands-on AEO walkthrough from a site that instrumented its own citability: the plumbing files, a zero-to-six Bing reading, an llms.txt server-log test.
- [Hooks, skills, subagents, plugins, CLAUDE.md: what each is for](https://okaneland.com/primer/claude-code-mechanisms-measured/): Claude Code's five extension mechanisms sorted by when they load into context: CLAUDE.md, skills, subagents, hooks, plugins. Token costs measured.
- [Ship an MCP server for your product](https://okaneland.com/primer/ship-an-mcp-server-for-your-product/): Build an MCP server under the 2026-07-28 spec, plus distribution receipts from @okaneland/mcp: the download curve, six directories, one listing still stuck.
- [How to build an app with AI from scratch](https://okaneland.com/primer/build-an-app-with-ai-from-scratch/): A map of the real path from idea to shipped app with AI: scope, tool choice, the 70% wall, deploy limits, security, and first users. Prices verified 2026-08-09.
- [Can you actually learn while vibe coding?](https://okaneland.com/primer/can-you-learn-while-vibe-coding/): Can you learn while vibe coding? Only by breaking its defining rule: read the code. The three moves that build real understanding from AI-written code.
- [How to use n8n, and when to write the code instead](https://okaneland.com/primer/how-to-use-n8n/): A practical guide to n8n: execution pricing vs Zapier tasks, self-hosting, AI agent nodes on a leash, real client rates, and the license trap for agencies.
- [How to use Lovable, and when to take the code and go](https://okaneland.com/primer/how-to-use-lovable/): A practical guide to Lovable: the plan-first workflow, the credit math, the security incidents, and when to export the repo and finish in Cursor.
- [How to make AI UGC ads, and when to pay a human instead](https://okaneland.com/primer/how-to-make-ai-ugc-ads/): A practical guide to AI UGC ads: the Arcads, Creatify, and HeyGen workflow, hook-matrix testing, the real credit math, and when to pay a human creator instead.
- [The vibe coding stack: what to pay for and what to skip](https://okaneland.com/primer/vibe-coding-stack/): The vibe coding stack, slot by slot: editor, agent, app builder. What each one costs a month, where the $20 plans really land, and when free is enough.
- [How to run Claude Code, and when to take back the keyboard](https://okaneland.com/primer/how-to-use-claude-code/): A practical guide to Claude Code: the explore-plan-implement-commit loop, the usage pool and weekly caps, and when to reach for Cursor instead.
- [Where Perplexity earns its keep, and where it doesn't](https://okaneland.com/primer/how-to-use-perplexity/): A practical guide to Perplexity: Pro Search vs Research mode, Focus and Spaces, the real pricing, and where you are better off with Google.
- [How to actually use Cursor, and where it stops being worth it](https://okaneland.com/primer/how-to-use-cursor/): A practical guide to driving Cursor's AI editor well: the four surfaces, the rules file, the pricing trap that burned its fans, and when to switch tools.

## The Studio: builder interviews
- [Studio 03 · Tom Huang, Open Design: "Stars are not revenue wearing a trench coat"](https://okaneland.com/studio/tom-huang/): Open Design shipped 11 days after Claude Design and hit 1M+ installs on 77k stars. Tom Huang on the file-first method, the $200 build, and why he gave it away.
- [Studio 02 · Gabriel Amzallag: "I only build it if someone pays"](https://okaneland.com/studio/gabriel-amzallag/): Gabriel Amzallag pre-sold Web Anatomy before building it: a library of what converts on landing pages, agent-written on a €150/month stack. First sale, €99.
- [Studio 01 · Steven Hylands: “Build judgement, not the thing you've already got”](https://okaneland.com/studio/steven-hylands/): Stora co-founder Steven Hylands on 10,000 pull requests, a quarter of them in the last six months, and why judgement is the one skill AI can't hand you.

## The Proof: hands-on tested reviews
- [Rank Prompt on the free plan: it scored our AI visibility at zero, and by hand we found citations on two of eight questions](https://okaneland.com/proof/rank-prompt/): Rank Prompt's free AI-visibility report, tested against a hand count on our own site. It said zero. By hand, citations on 2 of 8 questions. Both are right.
- [Chatbase on the free plan: we fed it our own site and asked it twenty questions we already knew the answers to](https://okaneland.com/proof/chatbase/): Chatbase's free plan, tested on our own site: 18 of 20 answers exact, none invented, one credit a message. The 1 MB cap dropped 22 pages, homepage included.
- [Google Antigravity at $0: we gave it the same booking app and it ran out of models by half past four](https://okaneland.com/proof/antigravity-free/): Google Antigravity's free tier, run through our frozen 12-gate booking-app test. Gemini, Claude and GPT-OSS all hit the weekly quota wall in one afternoon.
- [Codex on the ChatGPT Free plan: we gave it $0 and the same booking app](https://okaneland.com/proof/codex-free/): OpenAI Codex on the $0 ChatGPT plan, run through our frozen 12-gate booking-app test. 12 of 12, no Astra, and a 30-day meter that one build ate 37% of.
- [Web Anatomy: the annotations are worth your time, the statistics are not](https://okaneland.com/proof/web-anatomy/): We tested Web Anatomy's rubric for reproducibility, ran its analyzer on our own page, and checked its analyses against live sites. Worth it? Situational.
- [Cursor vs Claude Code for vibe coding: a real head-to-head](https://okaneland.com/proof/cursor-vs-claude-code/): Cursor Pro vs Claude Code across two unlike builds and 24 frozen gates. They tied both times, 23/24 each. What still separates the two $20 agents.
- [Cursor Pro: we gave it $20 and a startup idea](https://okaneland.com/proof/cursor/): We gave Cursor Pro one founder pitch and it built a working SaaS: named itself, 23 passing tests, CSV import, a freemium demo checkout. What $20 buys.
- [Open Design: the free design studio your coding agent runs](https://okaneland.com/proof/open-design/): We ran Open Design end to end with Claude Code as the engine: brand extraction from our own site, a component kit, a media kit, and the real token bill.
- [Creatify: the product-page ad machine, for the right seller](https://okaneland.com/proof/creatify/): We bought Creatify Pro and drove every tool: URL-to-Video, custom avatars, static ads, ad cloning. Situational: strong on retail listings, weak off them.
- [Clay: the best prospecting engine we have tested, and the bill that comes with it](https://okaneland.com/proof/clay/): We found 50 real leads, ran the six-provider email waterfall, and drove Clays AI agent live. The AI is real and the coverage is high. The catch is the price.
- [Perplexity Max: the deepest toolset in AI search, and where it still leaks](https://okaneland.com/proof/perplexity/): We bought Max and drove every tool: nine models, Model Council, Deep Research, the Computer agent, and the premium connectors. Powerful, but most need Pro.
- [Surfer SEO: a strong scoring engine wrapped in a paywall maze](https://okaneland.com/proof/surfer-seo/): We ran Surfer end to end on a live site, then checked our verdict against real users. The dual SEO and AI-Search scoring is real. The AI writer is not.

## Tools
- [The Palette](https://okaneland.com/palette/): a curated shortlist of AI tools worth your time, one line each, filterable by what they do and whether they run locally or in the cloud.
- [AI tool price index](https://okaneland.com/palette/ai-tool-price-index/): verified prices and unit costs for the AI tools we test, re-checked monthly, with raw JSON at https://okaneland.com/api/v1/prices.json.
- [AI margin calculator](https://okaneland.com/palette/keep/): what your MRR is really worth after the model bill, fees, churn, and tax.

## Link back
- [Link back](https://okaneland.com/link-back/): 88 by 31 buttons and the vibe coders webring.


## About
- [Media kit](https://okaneland.com/media-kit/): boilerplate, brand assets, colors, fonts, and editorial standards.
- [Privacy](https://okaneland.com/privacy/): what we collect and how to switch analytics off.

## Optional
- [Sitemap](https://okaneland.com/sitemap-index.xml): every indexable URL.
- [Full text for agents](https://okaneland.com/llms-full.txt): the full text of every published piece in one file.