Building SaaS with Claude AI: a category-by-category index

Most “build a SaaS with AI” tutorials stop at the demo screen. They show Claude generating a working signup form and call the job done. The hard part — the part that decides whether your product becomes a real business — happens after that: the multi-tenant data model, the webhook idempotency, the pricing tier that actually converts, the differentiation against an incumbent with a six-figure ad budget. The guides indexed on this page cover that part. They are written for solo founders who have already decided Claude is the build partner and now need to make 24 smaller decisions per project.

The pillar guide, how to build a SaaS with Claude, lays out a six-step framework: data model first, then auth and tenancy, then the core CRUD, then payments, then deploy, then growth scaffolding. Every vertical guide below follows that same skeleton, but swaps in the schema choices, edge cases, and prompt sequences that are specific to that product shape. A CRM is mostly a contacts graph with custom fields. An invoicing app is mostly PDF generation and Stripe Payment Links. A status page is mostly monitor scheduling and incident state machines. The shapes are different; the build flow is the same.

The guides also each end with a section most tutorials skip: the honest competitive picture. If you are building a notes app in 2026, you are competing against Notion. If you are building a feedback tool, you are competing against Canny. Each guide names the incumbent, the wedge that is still available, and the niche-down move that turns a commodity into a defensible micro-SaaS. That is the part that makes the difference between an interesting weekend project and something a customer will pay $29 per month for.

Business-operations SaaS is the inverse of productivity. The market is fragmented, customers expect to pay, and the willingness-to-pay is high because the tool sits next to revenue. A small business that does $500K a year in invoicing will pay $49 a month for an invoicing tool without thinking about it. The build complexity is higher — you are dealing with PDFs, accounting integrations, tax compliance, payment processors — but Claude handles most of it well because the patterns are public and well-documented. The four guides in this section cover the four most ship-able business-ops shapes: invoicing, CRM, booking, and job boards. Each one solves a real recurring problem for a specific kind of customer.

What to look for in each guide: the integration surface. The invoicing guide spends most of its prompt budget on Stripe Payment Links and PDF generation because those are the load-bearing pieces. The CRM guide handles the contacts-with-custom-fields data shape that determines whether you can serve a vertical niche. The booking guide handles conflict prevention and payment-on-booking. The job board guide handles SEO-friendly listing pages and the seed-then-convert distribution pattern that is the only known way to bootstrap a job board to revenue. Pick the integration surface that matches the customer you can already reach.

These are the build guides that lean into Claude as a product, not just as a build partner. The AI chatbot guide uses Claude (and the API token math from our cost calculator) to ship a doc-grounded retrieval widget. The receipt scanner uses Claude vision for OCR and structured extraction. The knowledge base uses a hybrid pgvector + Postgres full-text search index to serve grounded citations. The analytics dashboard handles a multi-tenant third-party API integration pattern that is reusable across half of all B2B SaaS. These are the highest-margin guides in the index if you nail the unit economics, and the lowest-margin if you do not.

The shared risk across the category is API cost. The chatbot and knowledge-base guides both spend serious time on the margin math, because at 100K user messages a month the token bill becomes meaningful. Each guide gives you the actual numbers and the four cost-cutting levers (prompt caching, smaller models for routing, batch where possible, retrieval limits per query). If you skip the cost section and just ship, you will discover at $3K MRR that your gross margin is 18%, and that is the failure mode that kills more AI SaaS than any product mistake. Read the cost sections carefully.

Once you have picked a vertical, the next problem is idea selection within that vertical. The micro-SaaS idea lists below are the source most of the build guides on this page borrowed their example customer profiles from. The “build a SaaS without coding” guide is the right starting point if you are not sure Claude (or any AI code generator) is the right level of technical investment for you yet. The AI-SaaS-ideas list focuses specifically on the AI-native category if you want to lean into that margin profile and are willing to pay close attention to the cost math.