GTM Jobs is a weekly email newsletter featuring hand-picked GTM opportunities across the US and Europe. Whether you’re looking for roles in GTM Engineering, Product Marketing, RevOps, Marketing Engineering, or AEO, you’ll find curated openings to help you land your next dream job.

The new roles companies are hiring for as they rebuild their Go-To-Market with the support of AI.

GTM Jobs

A weekly roundup of the hottest Go-To-Market jobs across the US and Europe.

By Stefanos Karakasis

Subscribe to get full access to the newsletter and publication archives. Upgrade to a paid subscription to support our efforts, join our Talent Community and get access to recruiters from companies with a solid working culture.

See you next week,

Stefanos

👋🏽 If you’re in Tech (PM or PMM) and looking for a new role, fill out this Candidate Form. We will review our list when helping place candidates directly with network companies and will r…

Read more

On building with AI

Describe the moment you realized your team's AI setup was actually getting in the way?

It was watching colleagues get stuck. Not because they lacked ambition—there was real willingness and urgency to use AI—but because the infrastructure just wasn’t there. Navigating access rights, multiple LLM options, MCP configurations... it was too much to grok. Some departments that pushed through had completely different setups, and there was no general way to share back what they’d learned. Everyone was solving the same problem in isolation.

I realized the bottleneck wasn’t technical capability—it was onboarding. So I shifted from automating individual Product Marketing (PMM) workflows to building an AI productivity suite for the department where configuration comes out of the box. Make it so people don’t have to figure it out from scratch.

Walk me through the first three things you built—and which one taught you that you'd been thinking about the problem wrong?

I started by using Claude Cowork to build a roadmap of which workflows to automate first. That was meta but useful—I was essentially asking Claude to help me think about where the real leverage was.

Second, I settled on core sales workflows: outreach, call preparation, call digests. Those are the jobs where I could see people drowning in repetitive work. I didn’t start from scratch—I took existing Claude Skills from Paweł Huryn, Corey Haines, and Anthropic’s own GitHub repo. The insight there was: don’t build, adapt. Look for what’s already working in the community and customize it for your context.

Third—and this is where I realized I’d been thinking about it wrong—I built a fully authenticated memory system into each Skill. Every session gets logged, and every 24 hours we run a synthesis pass. That’s when I understood the real problem wasn’t the Skill itself, it was retention. People were repeating context every session because there was nowhere to store it. Once I added the memory layer, everything else made sense.

The teaching moment: I’d been focused on what Skills do when I should have been focused on how people persist information across sessions. Without that, you’re building tools that help for an hour, not systems that compound.

What’s something you tried to generate end-to-end with AI that completely failed?

The enablement training for managers. I thought: “I’ll feed Claude the context about how these Skills work, the ethical implications, the change management aspects, and it’ll generate a training program.” The output looked polished. It was wrong.

AI can’t replace the conversation you need to have when you’re actually changing how work works for real humans. There were ethical questions baked into that training—around data access, autonomy, what it means to have an AI think with you every day—that required human judgment and dialogue. I tried to generate my way around that conversation and it showed.

That failure was important. It taught me the boundary: AI is incredible at synthesis, pattern-finding, scaling your thinking. It’s terrible at replacing the human ethics work. Don’t try.

Show me one actual prompt or system you built—walk me through why it's structured the way it is, not what it does?

I’ll use the Pipeline-Digest skill. The structure looks like this:

First, it doesn’t just fetch your Salesforce data and generate a report. It starts by asking three diagnostic questions:

• What’s your role?

• Which market are you managing?

• What’s your definition of “momentum” right now?

Why? Because I learned the hard way that if I just let Claude execute, it would hallucinate what mattered. It would optimize for the metric I thought was important, not the metric you actually care about. So I forced a calibration step before any execution.

Second, after it gets those answers, it doesn’t generate a generic weekly digest. It runs a synthesis loop: pull the deals, categorize them by momentum/at-risk/stalled, surface the patterns and then ask follow-up questions about why those patterns exist. It doesn’t conclude—it interrogates.

Third—and this is the retention piece—every run gets logged with your role, market, and calibration. When you run it next week, it remembers what mattered to you last week. It can then ask: “Last week you flagged these three deals as at-risk. What changed?” Instead of starting from zero, it’s building on prior context.

The structure is interrogation → execution → memory. Not just “here’s your answer.” That’s why it works as a system and not just a tool.

What's one thing you thought AI would solve that it actually didn't? Or made worse?

It didn’t push me proactively on revealing any blind spots in my requests. It just executed, and the output looked all good at first, but it didn’t help build the most scalable system from the outset. For anything that is going to be used daily, you need synthesis and consistent review of how Skills are performing. I learned this the hard way by adjusting my prompting to have it always ask questions before execution. Also tried a fully generative approach to creating the enablement training for managers—garbage. AI can’t replace the ethics conversation you need to have when you’re changing how work works for actual humans.

What’s one thing you learned from someone else that changed how you think about building with AI?

taught me about building review loops that audit the Skills themselves. The insight was: you can’t just build a Skill and launch it. You need to build a system that continuously reviews whether each Skill is performing, whether it’s structured consistently with the others, and whether it’s still calibrated to what people actually need.

That flipped my thinking. I was so focused on building the user experience—the Skills, the outputs, the integrations—that I wasn’t building the operating system that keeps all of it honest. Rick’s point was: the system is only as good as your ability to review what the system is doing.

Now, every Skill I build includes a review protocol. Not as an afterthought—as part of the architecture. That changes everything about how you think about scale.

Complete this sentence honestly: 'The mistake I made earliest was _____ and I didn't realize it until _____.

The mistake was assuming AI would tell me when something was broken. I’d build a Skill, it would execute perfectly, the output would look polished, and it wouldn’t actually be helping people.

I didn’t realize it until I started getting quiet feedback: “It’s nice, but I have to rewrite most of it anyway.” Or: “It assumes things about my role that don’t apply.” The Skill was working—it just wasn’t working for the actual person using it and their role.

That’s when I understood: AI doesn’t reveal your blind spots by default. It simply executes your assumptions and makes them appear credible. If you want to build something useful, you have to force the interrogation. That’s why I restructured everything to ask questions before execution, to run synthesis reviews on all the Skill outputs, and to build feedback loops into the system.

The blind spot wasn’t about AI. It was about me not having a process to surface when my assumptions were wrong. Once I established that process, everything improved.

Side Notes

If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already.

Upgrade to paid

👋🏽 If you’re in Tech (PM or PMM) and looking for a new role, fill out this Candidate Form. We will review our list when helping place candidates directly with network companies and will r…

Read more

👋🏽 If you’re in Tech (PM or PMM) and looking for a new role, fill out this Candidate Form. We will review our list when helping place candidates directly with network companies and will r…

Read more

👋🏽 If you’re in Tech (PM or PMM) and looking for a new role, fill out this Candidate Form. We will review our list when helping place candidates directly with network companies and will r…

Read more

Read more

👋🏽 If you’re in Tech (PM or PMM) and looking for a new role, fill out this Candidate Form. We will review our list when helping place candidates directly with network companies and will r…

Read more