The Pangram Labs study (published July 9, 2026) scanned over 1 million posts across five platforms. LinkedIn was the most AI-saturated by a wide margin: over 40% of long-form posts (250+ words) were fully AI-generated, and LinkedIn accounted for 62% of all AI-flagged content across the five platforms studied.

It’s not just LinkedIn; I stopped really investing in X when the algo update no longer rewarded the kind of content I enjoyed posting. We’ve recently seen our YouTube channel take a huge hit as the algorithm has changed to account for AI-generated videos.

And so, this month, I’ve seen a huge drop in the performance of my posts on LinkedIn, and I thought, huh, I wonder why. LinkedIn’s content analytics suck. So, I got busy and built my own dashboard featuring a new metric that I believe is the most important to track if you want to grow an audience on that platform.

IMPORTANT NOTE: People over on LinkedIn pointed out you’re not allowed to use browser extensions to scrape data from LinkedIn. My recommendation is not to use the Claude skill; instead, take screenshots of your posts’ analytics. Have Claude then parse the images for the metrics cited below. Will leave Claude skill below as can be adapted for other purposes.

In this post, I’ll walk you through it, and you can download a templatized version of the dashboard and a Claude Skill to build your own version.

See the live demo dashboard →

What You Get

A single interactive dashboard that lives inside Claude as an Artifact you can update anytime, populated by a single Claude skill that uses Claude computer.

The interactive dashboard features:

• Monthly trend charts for impressions, reach, followers, gained, and engagements.

• Click any month to see an overview of performance for that month, broken out at the post level.

• Top 10 leaderboards broken out by impressions, engagements, and virality.

• In-network vs. out-of-network reach so you can see which posts break out of your bubble. THIS is the most important metric for LinkedIn that no one tracks. The dashboard makes it easy.

• A 🔥 viral badge on posts where more than half the reach came from outside your network

• Full post content you can click to expand and re-read

A living dashboard inside Claude that updates every time you run the skill.

As per above, the following has been updated to account for the fact using Claude’s Chrome extension to gather data is against LinkedIn’s TOS.

Step 1: Screenshot Your LinkedIn Analytics

Go to your LinkedIn analytics page. You need to capture two things for each month you want to track:

Your post list. Open your analytics, filter by the month, and screenshot the table showing all your posts with their impressions, engagements, and other stats. If your posts don’t fit in one screenshot, take multiple ones. Get every post.

Individual post analytics. For your top-performing posts, click into the detailed analytics view and screenshot the breakdown showing in-network vs out-of-network reach, followers gained, and members reached.

Don’t overthink this. If you can see the numbers on your screen, Claude can read them.

Step 2: Build Your Dashboard

Open a new Cowork session in Claude. Upload all your screenshots and use this prompt:

“Extract all the LinkedIn analytics data from these screenshots. For each post, capture: the post content (or first line), date published, impressions, engagements, followers gained, members reached, and in-network vs out-of-network percentage. Then build me an interactive HTML dashboard as an Artifact with monthly trend charts, a leaderboard of top posts, and month-by-month post breakdowns. Flag any post where more than 50% of reach came from outside my network as viral.”

Claude will read every number from your screenshots, structure it all into data, and build the full dashboard. The whole process takes a couple of minutes.

Step 3: Keep It Updated

Next month, take fresh screenshots and open a new Cowork session. Upload them alongside your existing dashboard Artifact and say:

“Here are my LinkedIn analytics screenshots for August 2026. Add this month’s data to my existing dashboard and rebuild it.”

Each time you do this, your dashboard grows. More months, more data, more patterns.

The reason you’d want to go back and update previous months is that content on LinkedIn has a long tail. Old posts continue to gain impressions and engagement. I’d suggest always refreshing the previous month’s screenshots when you add a new one. When I populate August 2026, I’ll also re-screenshot July’s numbers to catch the late-arriving metrics.

How this helps you do better at LinkedIn

There is so much great analysis you can do with this dashboard, let me go through 3 great ones:

1. In-Network vs Out-of-Network Reach: I remember feeling like there was a similar drop in my posts performance around February. But, I wasn’t sure, until I saw this chart:

How much of your posts engagement are coming from within your followers (In) vs. outside (Out). The more coming from outside, the better you’ll perform. You’ll see July-26 is trending to be a poor month, similar to Feb-26. This tells me LinkedIn, for some reason, is showing my posts to less people from outside of my followers.

2. Chat with Live Artifact: Chat to your live artifact and gather insights about the data it’s showing you. For example, I want to know why July is performing so poorly on ‘In-Network vs. Out-of-Network” vs. June. After going back and forth it told me:

• My last 3 posts were to ‘inside baseball’ targeting my existing followers vs. people who’d sit outside my network. I loosely agree. I do believe LinkedIn changed something 7 days ago.

• It pulled out a couple of posts in June’26 that would appeal to a much broader audience and gave me ideas of experiments to run.

• It said to cut the self-promotion lol. I don’t believe this is accurate, it’s because in July I referenced things I built. It marks them as self-promotion.

It’s incredibly helpful to chat about the data and gather insights. The fact all post text is added to the interactive dashboard allows you to derive real insights from writing patterns that are working vs. not working.

3. Top Leaderboards: I love this tab, it allows you to see your top 10 posts across impressions, engagements, and virality (posts with > 50% out-of-network engagements). Add the chat feature and you can gather insights into what is working, not working and what’s changed.

The key to getting better at anything, including growing your audience on LinkedIn, starts with looking at the right data & insights.

» Get the skill and preview the dashboard

Until Next Time,

Happy AI’fying

Kieran

Not more productivity. More chaos. A tangled web of ‘AI Sprawl ‘.

Everyone turned into an AI builder overnight. Some did it because it was now a big part of how they got promoted. Some did it because their team was measuring usage, and they had to be seen using AI tools. And a lot of people did it because they realized building stuff with AI was way more fun than their actual job.

The C-suite mandated “use AI”. The org delivered chaos. Because AI without a system is just that - total chaos.

The numbers are brutal.

A recent Zapier survey found that 78% of employees using AI at work adopt tools outside IT approval. The IT Armageddon. 31% of enterprises discover new “rogue” AI tools monthly. And a recent report showed teams using more than 5 AI tools report lower self-rated productivity than teams using just 1-2 (Coommit).

In HubSpot, as we’ve dug into AI usage across the company, we’ve often found a negative correlation between people’s outcomes and their usage of AI vibe-coded tools they’ve built to help them do their work.

MIT research found that 95% of organisations see no measurable returns from AI investments in knowledge work. Gartner is forecasting that more than 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear value, and inadequate controls. And 54% of C-suite executives now admit AI adoption is, in their words, “tearing their company apart”.

Niranjan Vijayaragavan, Chief Product and Technology Officer at Nintex, called it plainly:

the rush into generative AI is creating “a new class of inefficiency.”

Companies invested in AI to move faster. The sprawl is making them slower.

But there is gold beyond that rainbow. There are companies and teams that are transforming how they work through AI. What are they doing differently?

From sprawl to system: agentic pods.

Uber recently shared how they’re approaching this. They call them Agentic Pods. The concept is simple: pair an AI-proficient engineer with a domain expert, give them a tight window, and ship a working end-to-end workflow.

Their results: marketing web QA went from 2 weeks to 50 minutes. Capital allocation across 150 cities: 15 hours to 30 minutes. Financial pacing reports: 2 days to 10 minutes.

This is the way.

We’re running a version of this at HubSpot right now. Engineers embedded with GTM teams, building workflows against real systems, measuring what moves the number.

But Uber’s model stops at “ship.” And anyone who’s actually done this at scale knows that shipping is maybe 40% of the job. The workflow has to survive contact with real users, hold up when you expand beyond the pilot group, and actually get adopted.

Here’s the full agentic loop we’re building toward:

1. Observe. Both qual and quant. If practical, e.g., you have budget, there is a new set of tools that will help you map workflows across your GTM teams and connect them to business outcomes Observability for GTM workflows. That means you can then go shadow the GTM practitioner/s doing those workflows. Map it out. See if it’s worth moving into the agentic bucket.

2. Identify. Find the highest-impact automation opportunities. Not every workflow is worth rebuilding. These need to be correlated to outcomes. You have to be able to measure what the impact of workflows is and prioritise, everything will look like an opportunity; this is the hardest part of the loop.

3. Build. Engineer and domain expert build the workflow together. Not for them. With them. To be honest, the domain expert can build most of it if using Codex, Cowork, etc. The engineer can build foundational features that might be needed to scale that workflow within these LLMs.

4. MVP. Deploy to a small group first. See if results hold before you roll it wide. It’s one thing getting a single person to use it and show impact; it’s a whole other getting 100s or 1000s.

5. Expand. Push it to more people. Does it generalize, or did it only work for the pilot group?

6. Enable. The most undervalued step. Someone needs to hold hands, train, and make sure the workflow actually gets used. Building it is maybe 40% of the job. Getting adoption is the other 60%. Changing human behavior is incredibly hard.

7. Measure. Close the loop back to observability. What worked, what didn’t, and what do you build next?

You’re likely thinking.

“This works if you have engineers to spare. Most GTM teams don’t.”

The pod model doesn’t require a full engineering team. It requires one technically proficient person paired with one domain expert. In some organizations, that technical person is a GTM engineer or just the person who is home every evening building out agents, deep in the AI weeds.

The point being, measuring “AI native” by usage is pointless. A person with low domain expertise and high AI agency is likely an incredible AI slop machine in your company right now, creating mass amounts of sprawl. You need to move from “everyone uses AI tools” to “let’s instrument the workflows that actually drive outcomes.”

Until Next Time,

Happy AI’fying

Kieran

But I still believe there is real value in AI for content, in its ability to build a profile of your audience, better understand that audience, research ideas, create first drafts, and review your content.

As I work on v2 of my AI brain and AI content system, I wanted to do something different in this post and share a skill from the upgraded version of the content system.

Can AI help you instill taste into the content you create by acting as a coach?

The skill is called content taste. It’s a check-in for content you’re going to publish. A system that asks the two questions most writers skip.

Gate 1: The Share Test.

Would your ideal writer forward this to a colleague with a note that says “you need to read this”?

People share content for exactly five reasons. It made them look smart. It named something they felt but couldn’t articulate. It gave them an unfair advantage. It surprised them. Or it told a story they couldn’t stop thinking about.

Most content fails because it’s trying to be “useful” when it should be targeting “shareable”. Your currency as a writer is shareable.

Gate 2: The Onlyness Test.

Do you have a differentiated perspective? Is your content unique? Do people know it’s yours?

This is where people have gone awry on AI. Their content is perfectly written, word by word, by a series of prompts and intelligent AI models, and all that content is indistinguishable from each other.

What’s unique is your specific opinion, your lived experience, your achievements, what you know.

The skill tests in three ways. Delete the author’s name. Can you still tell who wrote it? Could a competitor republish the same piece next week? Is this built on something only this creator has, or on the same public inputs every model has already seen?

It coaches and provides helpful recommendations.

The skill will provide suggestions on how to improve your content and

1. What do you believe about this that your smartest peers would argue with?

2. What did you personally watch happen that contradicts the consensus?

3. What’s the number, result, or story that only you have?

4. Who is this for, and who are you willing to be wrong in front of?

Unlike most skills, it doesn’t give you generic recommendations; it attempts to coach you on how to improve your content.

The anti-AI slop patterns

The skill will then attempt to fix all those AI ‘slop’ patterns that have become so common in online content. I should note that these patterns are not ‘slop’; they are great writing patterns; it’s just that AI has saturated them.

How is this giving me taste?

The purpose of this taste is to coach you on how you can develop the right instincts to make your content ‘tasteful’.

The purpose of the skill is to help force a conversation on how you can improve your content. It’s to help you pause and be a little more considerate of what you are posting and why. This is what I use the skill for; these words are as much for me as you.

Would someone actually share this? Could only I have written it? Is there a single line worth screenshotting?

You likely know the answers immediately. But, you rarely stop long enough to ask yourself them.

When something fails, it doesn’t attempt to rewrite your draft with better adjectives. It spars. Acts as a coach, pushing you to bring out your unique taste and inject it into the content.

One quick note. This skill is plugged into my system so auto loads my audience profile and writing style that I covered in this post on how you build. If you don’t have either of those things, it’s ok, I’ve modified the skill so it will still work.

The giveaway

So, as I said, this post is a little different, I wanted to give away one of the skills from the new content system I’ve been building to get a feel if you all think they are useful. Here is the downloadable SKILL.md. Here is the image, you can also just take this and ask AI to create it.

I ran one of my LinkedIn posts through it and this was the result, the spar means i can then go back and forth with the system to pull out things from me that would make the post better. This is the core difference in how it works with what i see a lot of other people posting. It’s still you making the content better, the AI is helping to direct you, acting as a sparring partner.

Feel free to drop me a note on how it works for you. As always, you should adjust to your needs.

I’ll be dropping much more from v2 Second Brain and v2 AI content system in future posts.

Until Then,

Happy AI’fying

Kieran

I’m one of them. So is Brian Halligan, the former CEO of HubSpot and now partner at Sequoia, who revealed he’d built an AI agent trained on his entire life that now attends his Zoom meeti…

Read more

These have been learned through years of grinding, and iterating to figure out how to scale a personal audience through content.

Where applicable, I’ve added in skills from …

Read more

The title of the post was “A frontier without an ecosystem is not stable.” It sounds like something written for a CTO.

But it’s as applicable for anyone working in marketing.

Buried in the post is the clearest description I’…

Read more

However, Fable 5 does deserve a different conversation. It’s a new class of model - the first Mythos-class model Anthropic has release…

Read more

Almost all of it is being written by a group of people who’ve already made their money, had great careers, and aren’t under the same pressures as people still trying to build one.

I want to tell you why this nar…

Read more

That’s a mistake.

If you integrate AI into how a team works, you’ll get minimal results. I believe you have to rethink how the team…

Read more

What shipped in the las…

Read more

We’ve seen this happen before.

When factories first got access to electricity in the late 1800s, most of them didn’t redesign their operations. They just replaced the steam engine wit…

Read more

I recently took over a group of 400+ people.

I’v…

Read more

“I’m totally overwhelmed by AI. I know you wouldn’t understand, but it’s how I feel.”

I do understand. Everyone working in the industry is overwhelmed. No matter what they’re posting about on…

Read more

Read more

The workers there were, by their own union’s admission, the worst in the American auto industry. Absenteeism ran at 20%. Cars rolled off the line with missing parts. Workers left empty bottles inside door panels to rattle and annoy future owners. GM closed it down.

Two years later, Toyota and GM re…

Read more

Having Ogilvy’s principles codified into a grader that scores your copy in real-time does.

Everyone consumes expert content. Newsletters, podcasts, YouTube. With AI, it’s now possible to instantly turn that content into action.

As Naval Ravikant put it:

“Reading is not the same as understanding. Understanding i…

Read more

When Pixar made Toy Story in 1995, they nearly didn’t finish it. The story kept breaking. Andrew Stanton, Pete D…

Read more

It’s not finished :)

But I’ve learned a lot.

I wanted to pause and list out some learnings f…

Read more

Read more

I say finished, but working with Claude Code is different. You’re never done. You’re constantly improving.

I’m publishing the full Content Team next week…

Read more