Case study · AI content platform for founders and creators

LinkedFlow

A full SaaS for writing, scheduling and publishing content across LinkedIn, Dev.to and Hashnode, with an AI writing studio, a profile optimiser, research tools and analytics. Designed, engineered and launched in-house, now in early access.

LinkedFlow homepage
Role
Product design, full-stack engineering, launch
Status
Live · early access
Stack
  • TanStack Start
  • React
  • TypeScript
  • Hono
  • tRPC
  • Drizzle
  • Cloudflare Workers
  • Cloudflare D1
  • Tiptap
  • Razorpay
  • Groq
  • OpenAI
  • Gemini
The problem

Why it needed to exist.

Founders who post consistently on LinkedIn grow an audience; almost none of them have the time. Writing, choosing a posting time, cross-posting a long-form piece to Dev.to and Hashnode, and then checking what worked means five tabs and an afternoon.

We wanted one place that takes a rough idea to a scheduled, multi-platform post in minutes, and then closes the loop with real analytics, without the tool itself becoming another job.

What we built

The product, feature by feature

  • Content studio

    A Tiptap-based editor with AI generation of posts, hooks, CTAs, hashtags and ideas, switching between Groq, OpenAI and Gemini behind one interface.

  • Scheduler with optimal times

    Queue posts against AI-recommended slots; a Workers cron publishes them and auto-posts the first comment.

  • Blog writer and cross-posting

    Long-form drafting that publishes to LinkedIn, Dev.to and Hashnode from a single source.

  • Analytics and reports

    Impressions, engagement and profile views pulled from the LinkedIn API, exportable as PDF, CSV or XLSX.

  • Profile optimiser

    AI audits of a LinkedIn profile with per-section scores and headline variants.

  • Research, connections, teams

    Web search with collections and alerts, a connections manager with recruiter detection, and seat-based team licensing.

LinkedFlow landing page, scrolled
LinkedFlow on a phone
How it's built

Engineering decisions worth mentioning

  • Typed end to end

    tRPC over Hono means the dashboard and the API share one set of types; a renamed field fails the build, not production.

  • Everything on the edge

    SSR frontend and API both run on Cloudflare Workers against D1, so there is no origin server to scale or pay for while idle.

  • OAuth done properly

    LinkedIn, Instagram and X sign-in with token refresh, so scheduled posts still publish weeks after the user last logged in.

  • Billing that survives webhooks

    Razorpay subscriptions with verified webhooks driving plan limits and seat counts.

  • 12

    dashboard modules, from content studio to team management

  • 3

    publishing targets from one draft: LinkedIn, Dev.to, Hashnode

  • 4

    AI providers behind a single generation interface

  • 0

    origin servers. SSR and API both run on Cloudflare Workers

Building something like this?

This is the same team, stack and cadence you'd get. Tell us what you're making and we'll scope it honestly.