An AI-Native SaaS — The Product, The Marketing, The Build Pipeline

The Ambition

We set out to build the first truly AI-native SaaS end-to-end. Not "a SaaS with a chatbot" — a system where every surface that touches the customer was produced by AI. The feature they use. The ad they see. The blog post that brought them. The landing page that converted them. The email that re-engaged them. Even the code that shipped it.

Plantory.ai is that system, live in production across eight countries.

AI Everywhere

AI in the product. A Gemini-powered advisor with full spatial context — the canvas, climate zone, soil, sun, and plant inventory. Photo-based plant recognition. Structured seasonal task generation. AI-rendered garden visualizations. Every AI call is grounded in the actual garden, not a loose text prompt.

AI in the media. A custom Satori + Resvg + Gemini pipeline generates every social post image, every SEO hero, every article cover. A Remotion-based video engine renders ads from AI-written scripts with AI voiceovers and AI visuals. Copy, layout, illustration — all machine-produced.

AI in the distribution. Campaigns are deployed programmatically to Meta Ads and Google Ads. AI builds creatives and targeting variants, sets budgets, launches flights, and reports on performance. A human sets strategy; AI runs the mechanics.

AI in the content. A weekly pipeline authors blog articles, newsletters, release notes, and programmatic SEO landing pages across eight locales. MDX publishing is automated end-to-end — translation, linking, metadata, images.

AI in the build. We created an internal Claude Code plugin marketplace — the plantory plugin with 20+ skills covering spec-driven development (/plantory:spec-specify, /plantory:spec-plan, /plantory:spec-breakdown, /plantory:spec-implement), GitHub Project orchestration (/plantory:board-seed, /plantory:board-ensure, /plantory:board-work), paid-performance reviews, founder LinkedIn workflows, Facebook trust-building, release pipelines, code review, refactor audits, and quality gates. The same Claude Cowork pattern we use for clients.

What's Actually in Production

End-to-end AI SaaS: onboarding, canvas editor, plant encyclopedia, streaming AI chat with image generation, task kanban, plant inventory, AI-generated shopping lists, photo markers with directional orientation, Stripe-backed billing with free tier and paid plans, eight-locale marketing site, programmatic SEO, automated social-asset generation, Remotion video ad pipelines, Meta + Google Ads deployment, blog + newsletter publishing, and a Claude Code plugin marketplace powering the entire build.

Tech Stack

• Product AI: Google Gemini (chat, vision, structured output) via the Vercel AI SDK — streaming endpoints, tool calling, evals, cost controls
• Media AI: Satori + Resvg + Gemini for still imagery, Remotion for video, AI voiceover for audio
• Ads AI: Meta Ads API + Google Ads API with AI-generated creatives, targeting, and budget allocation
• Content AI: MDX publishing pipeline with AI authoring, translation, SEO metadata, and automatic linking across eight locales
• Dev AI: Claude Code + custom plantory plugin marketplace (20+ /plantory:* skills), Claude Cowork patterns, spec-driven development
• Foundation: Next.js 16 (App Router), React 19, Tailwind 4, NestJS 11, Prisma 7, PostgreSQL, Redis, S3/MinIO, Stripe, Better Auth, Resend, Turborepo

Why It Matters for Clients

Every pattern we hand an enterprise — streaming AI endpoints, structured generation, multimodal vision, prompt engineering, evals, caching, cost controls, ad-platform automation, AI content pipelines, AI-assisted development with custom Claude Code plugins — we've already proven on our own time, our own stack, and our own budget.

When DX Heroes helps a company go AI-native, we don't hand them a deck. We hand them the playbook we run ourselves.

Read More

Three companion articles go deeper on Plantory from different angles:

• Why We Built Plantory — And What It Taught Us About AI — the founder narrative and what changed for DX Heroes
• The AI-Native SaaS Playbook — How We Built Plantory End-to-End — the architecture, the five AI surfaces, the stack
• Seven Lessons Plantory Taught Us About AI in Production — concrete, tactical learnings from running a real AI product