The $13/mo stack that replaces $542/mo of SaaS

Why $13 actually covers it: the API stack underneath

Every SaaS is a UI plus brand plus support layer wrapped around a tiny set of commodity APIs. The APIs themselves (database, email, payments, hosting, AI) are sold by infrastructure providers at near-cost, because they compete with each other on developer mindshare instead of marketing budgets.

When you pay HubSpot $50 a month, you're really paying for four things:

The "UI tax" (the gap between commodity API cost and SaaS price) was historically justified because most buyers couldn't access the API directly. Wiring a database to an email sender to a booking page used to require an engineer, and engineers are expensive, so paying $50 a month for someone else's pre-wired UI made sense. That assumption broke when AI coding tools collapsed the access barrier. Claude Code can wire Neon, Resend, Stripe, and the Google Calendar API into a working CRM with email and booking in an afternoon. The skill barrier that justified the SaaS markup is gone.

The API stack underneath, in the same language engineers use:

The paradigm shift, in one sentence: SaaS is renting software priced per-seat. The API stack is paying for the actual compute you used, priced at infrastructure cost. The ratio is usually ten to fifty times.

The same pattern shows up in every category:

The infrastructure providers (Neon, Resend, Stripe, Netlify, Anthropic) sell the same primitives that power HubSpot, Klaviyo, Calendly, and Webflow underneath, directly to you, at cost. The SaaS layer was never doing the hard part. It was selling you access to plumbing that's now available without a salesperson in the middle.

You only rebuild the 10% of features you actually use

The cost story is half the picture. The other half is that every SaaS tool is loaded with features you'll never touch, and you're paying for 100% of the platform to access the 10% of it you actually use.

Open HubSpot for the first time and you'll see workflows, sequences, deals, tickets, conversations, lists, marketing emails, transactional emails, ads, integrations, dashboards, reports, custom objects, automation, AI assistants, and a settings menu with forty-seven sub-pages. You'll probably use four of those: contacts, a notes field, one or two pipeline stages, and a list of who hasn't replied in two weeks. The other forty-three menu items are noise you have to navigate around every time you open the tool.

The pattern repeats everywhere. Most solo founders use maybe 10% of Notion, 15% of Canva, 5% of Klaviyo, 20% of Calendly. But you pay for 100% of all of it, every month. You also spend hours learning each platform's idea of how the workflow should work, which is shaped by a product manager designing for the average of ten thousand customers, not for you. Every tool has its own mental model, its own keyboard shortcuts, its own settings page. Switching between five of them all day is its own kind of tax.

When you build the same 10% yourself, you get three things SaaS can't give you:

This is the second reason the in-house stack feels lighter than it should. Not just because it costs $13 instead of $542, but because the cognitive overhead of running it is smaller, since every line of it exists because you needed it. The off-grid stack builder skill is the workflow for finding that 10% inside each of your current subscriptions, so the rebuild is targeted instead of a full reimplementation.

Audit your subscriptions before you cancel anything

Before you cancel anything, you need to know exactly what you're paying for and — more importantly — which features inside each tool you actually use. Most subscriptions are 100% of a price tag for about 20% of the features. The off-grid stack builder skill takes your subscription list (or a screenshot of your card statement), walks tool-by-tool, and asks the same two questions for each one: which features do you actually use, and how often. The first time I ran it on my own stack the total surprised me. Subscriptions stack up the way pocket change does. You only notice when you put them on a single page.

The audit is also where you find the dead weight. Most stacks have at least one tool that's still billing every month for something nobody on the team actually uses. Catching those is the first easy win. After that, you start ranking the rest by 'how much of my workflow does this actually own' and you'll usually find that two or three tools do most of the work and the rest are convenience layers you could happily replace.

If you're auditing a prospect or a client's stack instead of your own, the SaaS stack scanner is the B2B variant — it scans a company's public website to detect what tools they're paying for, so you can pitch them on the same off-grid logic with their specific stack in the message.

Decide what to build, what to keep, and what to drop

Not everything in your stack is worth replacing. The rule I use is this: if a tool is genuinely commodity infrastructure (the email delivery layer, the database hosting, the payment processor, the file storage) keep it. Those services do their one job exceptionally well, the per-usage pricing is fair, and rebuilding them yourself would be a distraction. Resend, Neon, Stripe, Netlify, and Cloudflare fall in this category.

If a tool is a UI wrapped around a database that you're paying $50 to $200 a month for (CRM, proposal builder, scheduling, design canvas, sequence engine, analytics dashboard), that's where the build-it case becomes overwhelming. The features you actually use are usually a small fraction of what you're paying for, and the in-house version can be exactly tuned to your workflow instead of someone else's idea of an average customer.

If a tool is a content workflow that you reach for daily (the design tool you make graphics in, the writing surface you draft posts in), the decision is more personal. Some of those tools have real product value beyond the database underneath. Decide case by case, but be honest about whether you're paying for genuine craft or just paying for habit.

The five tools to rebuild: landing page, leads, CRM, proposals, emails

The operational stack for a small business has five pieces that show up in basically every business I've built. Replace them in this order and the cancellation chain becomes easy.

1. The landing page (replaces Webflow, Squarespace, and a $30/mo bill)

A clean, fast, SEO-friendly marketing site, built as static files and hosted free on Netlify. The AI website cloning skill takes a reference site you admire and produces a clean version of your own site in an afternoon. You describe the structure (hero, what we do, how it works, social proof, pricing, FAQ, CTA), point at a style you like, and the AI produces the HTML and CSS. The result loads in under a second, ranks better than the typical SaaS site builder, and costs nothing to host.

2. The lead pipeline (replaces Apollo, Clay, and a $50-$150/mo bill)

The lead pipeline is the thing that finds your potential customers and gets them into your CRM. The lead-gen Apify skill uses Apify's scraper actors to pull lead lists from LinkedIn, Google Maps, and similar sources at about a tenth of the price of the SaaS equivalents. Results land directly in your CRM with the right enrichment fields, so the downstream outreach can pick them up without an export step.

The broader lead generation skill wraps the scraping, enrichment, and ICP scoring into a single workflow. Once it's set up, finding a fresh batch of 100 well-qualified leads in your niche is a five-minute job instead of a half-day of switching between four tools.

3. The CRM (replaces HubSpot and a $50/mo bill)

The CRM is the heart of the whole stack, and the one piece most founders dread building themselves. The good news is that the core of a useful CRM is small: a contacts table, an events table, and a dashboard that shows you who's owed a follow-up and which deals are warm. The full SoloStack template ships with this already wired up, but if you're building from scratch, the prompt is roughly 'create a contacts table with name, email, company, source, ICP score, last contacted; create an events table that links to contacts; build a simple web dashboard that lists contacts and shows their event history'. A weekend gets you a CRM that's better than HubSpot at the one job you actually need it to do.

4. The proposal and document flow (replaces PandaDoc and a $25/mo bill)

Once you're sending offers, you need a clean way to render a proposal as a shareable page or PDF and to take a signature. The SoloStack template includes a proposal flow with e-sign and an audit trail, but the pattern is generic enough to recreate: a templated proposal page that pulls data from the CRM, a public link the client can view, and a small signature widget. The AI can build this in a day and the running cost is zero beyond the database.

5. The email sequences (replaces Klaviyo and a $100/mo bill)

Email sequences are the most over-priced category in the SaaS stack relative to what they actually do. The mechanics are simple: a list of templated emails, a trigger that drops a contact into a sequence, a schedule that sends each email in turn, and a way to stop sending if they reply or convert. The Resend API handles the actual delivery for $10 a month. Everything else is a few hundred lines of code that the AI will write for you in an afternoon. The version you build can be far smarter about your specific customer journey than any generic SaaS template.

Pick which tool to replace first

Don't try to build all five pieces in one week. Pick the order based on what your current week looks like.

If you're already running a business and the most painful tool is the most expensive one, replace that first. Klaviyo and HubSpot are usually the biggest wins by dollar value. If you're starting from zero, build the landing page and the CRM first (those two unlock everything else), then add the lead pipeline once you have an offer worth sending leads to, then the email sequences once you have customers worth nurturing.

Build one piece at a time. Run it for a week. Make sure it actually works in production before you cancel the SaaS equivalent. The temptation is to flip everything in a weekend. Resist that. The cost of cancelling something and finding out a week later that your replacement is missing a feature you needed is way higher than the cost of paying for both for a month.

Use proper specs so the AI builds the right thing

A common failure mode is to ask the AI to build something big with a one-sentence prompt and end up with a half-broken result. The fix is to spend an hour writing a proper spec before you start the build.

The PRD generator skill turns a rough idea ('I want a CRM that does X, Y, and Z') into a proper product requirements document: user stories, data model, screens, edge cases, what's in scope, what's out of scope. The AI uses that document as its brief, which means the result you get back actually matches what you had in your head.

For more technical pieces (data migrations, API integrations, webhook flows), the technical spec writer skill produces a tighter document with the architecture decisions written down. That's the brief you hand to Claude Code when you want the implementation to be careful and intentional rather than improvised.

Sequence the work so it actually ships

Once you have the spec, sequence the work so it builds in the right order. Two skills help here.

The roadmap builder skill takes your PRDs and proposes a sequence: what to ship first, what depends on what, where the risk lives, and what the rough timeline looks like. For a solo founder, the value of the roadmap isn't pretty Gantt charts. It's the discipline of writing down 'these three things ship this month' and then actually holding yourself to it.

The prioritization engine skill handles the harder question of which features to build at all. It takes your list of candidate features, scores each against impact, effort, and confidence, and ranks them. You'd be surprised how often the feature you thought was urgent slides to the bottom once you score it honestly. The discipline of prioritising on paper is what stops the side-project death spiral of building everything and shipping nothing.

What the cancellation week looks like

Once you've built the replacement and run it in parallel for a week, cancellation is mostly clerical. Log into each SaaS tool. Export anything you'll need (contact lists, email templates, proposal templates, calendar settings). Move the data into your in-house version. Cancel the subscription. Update any links or integrations that point at the old tool. Move on.

The first cancellation feels strange. You've trusted that vendor for a year. By the third cancellation it starts to feel routine. By the fifth you're checking your credit card statement to see what else you can axe. That shift in mindset (from 'what tool should I buy to solve this' to 'I'll just build it') is the long-term unlock. The dollars saved are real, but the bigger win is the freedom.

What's next in the series

You've cancelled the SaaS, built the core stack, and you're running on $13 a month. The next article in the series is about how to get customers finding you — without paying for ads: four engines that bring buyers to you while you sleep, instead of you chasing them.

Skills referenced in this article

Every AI skill linked above, in one table you can bookmark.