In partnership with
Hi y’all,
I post on LinkedIn about 5x a week. 3,000,000+ impressions over the lasts 18 months. Writing and designing is not what breaks me. I actually enjoy the process.
The ideation & research is what kills me.
Every Monday morning since September 2024, I sat down and faced the same question: what the f*ck do I write about this week? I would scroll through saved articles, half-remembered conversations, and bookmarks I had forgotten making.
So I built one AI agent to fix this issue once and for all. It was my first AI agent I built that added real value to my day to day, and it was not nearly as complex as I expected.
Today I will walk you through exactly what I built it and how you can build one for your worst recurring bottleneck.
Did someone forward this email to you? Click here to subscribe so you don't miss out on future issues.
The Start
I opened Claude Cowork and typed this…
Eat that you fancy prompt engineers.
Claude started asking me a lot of questions. What sources do you want me to find? What topics does your audience care about? What does a good research output look like? What do you actually want out of this?
I answered each question out loud using Wispr Flow, a voice dictation app that types as you speak (highly recommend it). I also shared my LinkedIn profile so it could scan through my existing content and data.
We went back and forth several times. By the end, a working agent existed. I had not written a single line of code. Here is the full system below we ended up with.
My Full LinkedIn Content System
The diagram shows 5 steps: Research, Write, Review, Publish, and Learn. Let’s go through it.
Blue steps in the diagram run automatically. Green steps require me. The agents handle research and performance analysis. I handle writing and publishing. The parts that make my work mine.
The orange loop at the bottom is probably the coolest part I am super proud of building. Every two weeks, the agent reads my LinkedIn performance data and feeds metrics & insights back into Research phase. What performed gets more weight. What flopped gets cut. The loop turns a tool into a system that improves over time.
Now, every morning at 7am, a research briefing lands in my Gmail.
Market news, industry reports, and content ideas pulled from 40+ sources: Reddit, newsletters, VC reports, LinkedIn, X, and more. It’s my own curated newsletter I enjoy reading with my coffee.
When something catches my eye, I reply with what I liked and possible content angles. Those responses get automated and my agent drafts 2-3 content outlines to a Notion page automatically.
By Monday, a week's worth of vetted ideas and draft outlines already waiting for me. No blank page. Just pick, review, and start writing. Here is exactly what it looks like.
My Automated Notion LinkedIn Content Draft Templates
6 Rules To Start AI Automation
I have spent 200+ hours inside Claude over the last four weeks. Here is what I wish I knew on day one.
Question whether a task is even worth automating. If you do something once a quarter, skip it. If it eats an hour every week, automate it. I have caught myself creating solutions to problems I don’t really have. Ask yourself this before you build anything.
Break your task into steps. Take something like responding to a client RFP. That feels like one task. It is actually many: research the prospect, parse the ask, pull case studies, draft the summary, draft the approach, check it against their criteria. Most tasks are five or six steps disguised as one. Write them out. The first step that only needs your time and not your judgment is where you build.
Start small and stack. Build one step. Get it working. Add the next. My system started as a research agent. Everything else came weeks later.
When you get stuck, show it. Screenshot the error, tell Claude what you expected to happen, and what actually happened. Three sentences that save you an hour of guessing.
Use the right model for the job. Not every step needs your most capable model. Reserve it for the steps that require judgment: planning, research, complex reasoning. Use a faster model for the steps that just need execution: formatting, repeating, summarizing. The difference compounds across a week.
Save your context before you walk away. Ask Claude to write a summary of what was built, what works, and what is left to do. Delete old files and context you no longer need. Next session, you pick up exactly where you left off.
Solving this one problem changed how every week starts for me. You have a version of it. That one task that drains you every week. You already know which task it is. Start there.
'Til next time,
--Ali
P.S. A handful of you asked for the Ikigai template from last issue. Rather giving you a manual template to fill out, I built a badass web tool instead. Try it. Share it. It might just reignite your life’s purpose. Click the image below to begin.
About Me: I am Ali. 18 years building and advising B2B companies. Proud Houstonian. Astros season just started and we are looking good. Currently dabbling in AI to help growth-stage B2B companies not get left behind.
Market Volatility Exposes Weak Delegation
When markets get shaky, advisors don’t just manage portfolios. They manage fear, questions, follow-up and a flood of client communication.
That’s where weak delegation gets expensive.
If meeting prep, paperwork, CRM updates and account admin still run through you, response times slip and the client experience takes the hit.
BELAY created the free Financial Advisor’s Delegation Guide to help you identify what to hand off, what to keep and how to stay client-facing without losing control.
Inside, you’ll learn how to reduce bottlenecks, protect responsiveness and free up more time for the work only you should be doing.
