dailyPost

Inspiration

I am a LinkedIn ghostwriter. I help founders and creators write content that builds trust and gets them seen.

After working with a bunch of clients, I started noticing something. The hardest part was not figuring out what to say. It was showing up consistently.

I felt the same way with my own content. Some days I was on it. Some days… nothing.

That is what inspired this. I wanted to build something simple that helps people stay visible on LinkedIn without overthinking it.

What it does

dailyPost sends you one AI-written LinkedIn post every day based on your content strategy.

It looks at your goals, audience, and topics, then writes a post that fits your style.

The post shows up in your inbox every morning. You just copy it, paste it, and stay consistent.

No stress. No ghosting the feed.

How we built it

I used Bolt, Supabase, OpenAI, Resend, Make.com, and Netlify.

Supabase handles the backend, the database, and the edge functions.

OpenAI writes the post based on your strategy.

Resend sends the post to your email.

Make.com runs the daily automation that triggers the post every morning.

Netlify hosts the frontend and also handles the custom domain.

At first, I tried buying the domain through IONOS with Entri but kept getting blocked. My card kept getting declined. I kept getting errors and nothing was working.

So I switched to Netlify. Bought the domain there and hosted the frontend there too. It was super quick and seamless. No issues at all.

Everything else runs serverless on Supabase Edge Functions. The daily automation with Make is simple but reliable.

Challenges we ran into

The hardest problem was getting the AI to actually write good posts. Not robotic. Not generic. Not boring LinkedIn fluff.

I also tried using the OpenAI Assistants API because the writing quality was better. But the thread handling was buggy and unreliable. I had to switch back to chat completions because it was way more stable.

Another challenge was finding the balance between features and reliability. Some things had to wait so the core product could work well.

Accomplishments that we're proud of

I shipped something real. Something that works.

It is simple, stable, and does exactly what it is supposed to do.

One post.

Every day.

In your inbox.

I am also proud of how clean the backend is. No extra features. No clutter. Just the stuff that matters.

What we learned

Writing a good AI prompt is harder than writing code.

AI will ignore instructions unless you force it to follow very clear rules.

Simple is better. Fewer moving parts means fewer problems.

People do not need perfect. They need reliable.

Consistency is more valuable than anything else.

What's next for dailyPost

There are a few things I want to improve after the hackathon.

Move the daily automation from Make.com into the backend.

Add a better way for the AI to match different writing styles using example posts.

Make the post quality even tighter by improving the prompt.

Maybe add a dashboard where users can view and tweak their content strategy.

Right now, the writing quality isn't where I want it to be. It's pretty rough. But I'm working on making it a lot better by using the Assistants API trained on hundreds or even thousands of LinkedIn posts

This is just an MVP

But for now, the core works. One post. Every day. In your inbox.

That is the whole point.