Tom Johnson

The Beating Heart of Observability Why the OpenTelemetry Collector makes or breaks your framework.

beyondruntime.substack.com/p/the-beatin...

‣ Shape your data early.
‣ Prioritize security.
‣ Be deliberate with receivers.
‣ Export with efficiency.
‣ Monitor the Collector itself.

The lesson I keep coming back to is simple: an observability framework is only as strong as its Collector configuration.

I’ve spent the better part of the past year working with the OpenTelemetry Collector to ensure our full-stack session recordings include automatically correlated backend traces.

Here are the lessons I’ve learned (sometimes the hard way) about configuring the Collector 👇

ZurichJS Meetup: 1st Anniversary Special Edition | ZurichJS Before the winter code freeze sets in and the only thing you want to debug is your holiday light setup, it's time for on...

He's giving a talk at ZurichJS next week.

If you're in town, I recommend checking out their EOY meetup on Thursday 13 Nov, 18:00 CEST. 👇

zurichjs.com/events/zuric...

👇 @farisaziz12.bsky.social describes the pain of customer support perfectly.

It’s always exciting to see Multiplayer show up in real-world stories like this, as part of how engineers actually solve problems. Seeing it used to cut through the “screenshot chaos” is exactly why we built it.

How I Finally Made AI Useful for Debugging Over the past year, I’ve spent some weekends experimenting with different AI-powered IDEs and assistants, just vibe coding for fun.

Full article: mayanksharmasharma77.substack.com/p/how-i-fina...

"Effective debugging isn’t about speed. It’s about visibility and understanding. When AI has access to complete context, it becomes a real collaborator in that process."

👆 This is exactly why we built Multiplayer. 🤩

A 5-Minute Fix to Our CI/CD Pipeline Learn how a small CI/CD pipeline tweak cut 30 minutes off every build, saving hours of developer time and reducing GitHub Actions costs for a busy startup.

Full write-up: dzone.com/articles/fiv...

Optimizations don’t have to be flashy or complex and a 5-minute fix to our CI/CD pipeline saved us 5hr a day.

This is a reminder that even the most obvious optimizations can hide in plain sight when you’re heads down building the next big thing.

MCP Demo Night: Dev Tools at the Edge of AI · Luma We’re bringing together MCP builders to showcase their work. Join us for a live demo night to see how these leading teams are turning MCPs from concept to…

Save your spot: luma.com/joouzfzw

A sneak peak of my presentation for tomorrow’s MCP demo night 👀

If you’re in New York - come say hi!

I bet a Multiplayer full stack session recording + Claude Code would have caught that 😉

Building agents with the Claude Agent SDK How to get started with the Claude Agent SDK and best practices for using it.

www.anthropic.com/engineering/...

Claude Code error or human error?

From the latest Anthropic blog post: variable names don’t match (red)…misspelling of urgent (blue) … unnecessary second check (second yellow line)

What about 'workslop'? That one would count too 😅

🧵 What is an “AI Engineer,” really?

It’s one of the hottest job titles of 2025 but also one of the most misunderstood. Let’s unpack what the role actually means (and why it matters).

When your AI confidently ‘fixes’ production.
Don’t worry, I’m sure it learned from this. 😅

Sometimes the frontend data isn’t enough.

Sometimes (okay, always) you also want to know what happened in the backend.

How much time do you have? I feel like the risk zone is anywhere between 3-6 hours.

This is a good time to remind everyone of the AI Darwin Awards. 😆

Devs: I’ll just make a small change.

QA tickets: 🐶🐶🐶🐶🐶🐶🐶🐶

Start simple, release gradually, and let user feedback guide you. Less is more when it comes to MCP.

❓ Curious what others are seeing: what’s the most *useful* MCP tool you’ve come across so far?

For us, that meant focusing on two high-value use cases:

1️⃣ Fixing bugs (where we can pipe full-stack session data directly into an AI tool)

2️⃣ Building features (where annotations/sketches from a session replay add the needed context to AI prompts).

Don’t just map your API 1:1 into MCP tools. That creates context bloat, and LLMs aren’t great at wiring together dozens of endpoints. Instead, scope tools tightly around developer intent.

MCP servers are everywhere right now. But most are collecting dust.

The key lesson we’ve learned at Multiplayer: scope matters. 🧵

I repeat. DON’T UPVOTE. I don’t care about that.

I just want to hear your feedback:

👉 Would you use this mainly for debugging, testing, or feature development?

👉 Have you tried session replays before? What worked, what didn’t?

I repeat. DON’T UPVOTE. I don’t care about that.

I just want to hear your feedback:

👉 Would you use this mainly for debugging, testing, or feature development?

👉 Have you tried session replays before? What worked, what didn’t?

But I also know that “I promise it's better” isn’t always enough when you’re busy and already juggling priorities.

So I’d love to hear: what’s made *you* drop a tool that was working and try a new one? And what made the switch worth it?

Is it word of mouth, seeing a demo, hitting a pain point one too many times, or just plain curiosity?

From my side: I genuinely believe we’re building something that saves time, reduces context switching, and brings all your data into one place.

I build developer tools for a living, and I’ve been wondering about this a lot: once you have a workflow that “works well enough,” what’s the trigger to get you to switch to something different and/or (possibly) better?