New Post

Section 1: Behind the Scenes of My Week

Writing, Outreach Experiments, and the Fog of Focus

This week was a mixed bag.

On LinkedIn, I stayed mostly consistent, posting twice a day except for a couple of slower days. But on X, I felt the fatigue. I tried to keep up with at least one post a day, but a few cheat days crept in. Nothing stood out engagement-wise, but I’m reminding myself that consistency builds the base. Virality is optional.

The more interesting part of the week wasn’t writing. It was outreach.

I started DMing podcast creators across X, Reddit, and LinkedIn. The reason? A small problem I noticed while browsing podcast tools. Most AI tools focus on turning long-form podcasts into short video clips. But there aren’t many tools that turn podcasts into high-context written content like X posts, LinkedIn posts, or carousels.

A few podcasters responded with interest. Enough for me to keep exploring for 2 to 3 more weeks. I did find a tool called Podsqueeze doing something adjacent, so now I’m studying it closely to see if there’s still a real gap worth building for.

In the background though, focus has been hard.

One part of me wants to write and share. Another wants to cold DM and validate. Some days I’m buried in community forums, hunting for that one real micro-SaaS problem.

Still learning how to choose the next best move, not just the next available one.

Section 2: AI Workflow of the Week

Voice-Powered Marketing Assistant using OpenAI + ElevenLabs

This is one of the most complex but scalable workflows I’ve built so far.

The idea: instead of just generating marketing content with ChatGPT, what if we could plug in multiple AI agents to do everything from blog generation to image creation to distribution all triggered by voice?

What it does:

→ Accepts a voice or text input via webhook

→ Converts it into a blog post, image prompt, and image search query

→ Generates or finds relevant images, edits them if needed

→ Uploads to Google Drive, logs everything in Google Sheets

→ Sends final output to a Telegram channel

All of this happens in a single workflow, orchestrated using LangChain and n8n.

Where it shines:

→ Uses ElevenLabs for voice input

→ OpenAI agents for blog writing and image prompt crafting

→ Image generation + search using external APIs

→ Delivery to both internal (Google Drive, Sheets) and external (Telegram) destinations

Why this matters:

→ Many founders think in voice (while walking, commuting, or thinking aloud)

→ This lets you say your idea once and get a publishable asset in return

→ It’s not just AI automation — it’s agent collaboration, end-to-end

I call it: From Voice to Visuals

And it’s surprisingly robust.

🔧 Built with: n8n, OpenAI, ElevenLabs, LangChain

📦 Includes: 30+ nodes, 6 AI agents, 4 integrations

🔗 View full workflow and documentation

Section 3: Cold Outreach Insights

Quick context:

This week I tried something new.

Instead of pitching a product, I started with a problem.

I noticed podcasters have a bunch of tools to turn long videos into short clips.

But there’s barely anything that helps turn audio-only podcasts into solid written content (like X posts, carousels, or LinkedIn summaries).

So I began validating this:

→ Reached out to podcast hosts on X, Reddit, and LinkedIn

→ Had a few early conversations

→ Learned that some are interested, but the use cases are very narrow

→ Discovered Podsqueeze (a similar tool) midway and plan to dig deeper next week

No landing page. No fancy demo.

Just cold DMs, a shared Airtable, and some honest conversations.

Not rushing. Giving it 2–3 weeks before deciding if it’s worth building.

Takeaway:

Sometimes the best way to validate is to show up like a curious peer, not a salesperson.

Section 4: Personal Reflection

This week felt… scattered.

Some days I was deep in outreach

Other days I got pulled into writing or browsing forums for micro-SaaS ideas

It’s not that I wasn’t working

It’s just that I kept switching gears

The hard part isn’t execution

It’s sticking to one direction long enough to see results

And I’m still figuring that out

But I’m also learning to be okay with it

Because this is part of the solo builder process you don’t get perfect clarity on day one

You stumble forward, test things, eliminate dead-ends

This week reminded me of that

Section 6: New AI Tool I’m Watching

Podsqueeze – for podcasters who write

While researching tools for pod-to-text workflows, I came across Podsqueeze.

It takes a podcast episode and auto-generates:

→ Show notes

→ Timestamps

→ Blog post drafts

→ LinkedIn & X snippets

→ Newsletter blurbs

Feels like a calm assistant built for repurposing audio, without flashy gimmicks.

I’m not affiliated, just studying how they approached a very specific problem.

Still validating if there’s space for a variant focused only on short-form posts (X threads, carousels, etc.)

But if you’re running a podcast, this one’s worth a try.

→ Try Podsqueeze: podsqueeze.com