The GTM Engineer role is replacing entire ops teams in 2026.
What a GTM Engineer does:
> Builds lead enrichment pipelines (not clicks through Clay manually)
> Writes ICP scoring algorithms (not guesses at targeting)
> Automates reply routing and classification (not checks inboxes on a schedule)
> Monitors campaign performance programmatically (not reads dashboards weekly)
> Integrates every tool via API (not exports CSVs between platforms)
What they DON'T do:
> Write cold email copy (that's the strategist)
> Handle sales calls (that's the closer)
> Manage client relationships (that's the AM)
One GTM Engineer with Claude Code replaces 3-4 ops people doing manual enrichment, list building, and campaign monitoring.
The role didn't exist 2 years ago. By 2027, every outbound team over $3M ARR will have one. The companies that hire early will have 6-12 months of infrastructure advantage.
I used to think MCPs were the future of AI tooling.
Now I'm ripping them out and replacing them with CLIs.
MCPs are great in theory. They make it easy to connect to any tool you need.
You should definitely start there if you're completely new to working with agents.
But you'll notice problems the more you use them:
- Agents wandering through massive APIs trying to figure things out
- Huge token counts before any real work started
- Inconsistent results from the same workflows
To get around that, now I'm building CLIs instead. Deterministic and verifiable tools that work the same way every time.
Minimal tokens and the same outcome every time you run them.
And because they are code, you can test them to guarantee they get the right result.
Turns out I'm not the only one seeing this:
Dev Browser vs Playwright MCP:
- CLI: $0.88, 29 turns
- MCP: $1.45, 51 turns
- 39% cheaper, 43% fewer turns
Streamlinear vs Official Linear MCP:
- Official MCP: 19,659 tokens
- Streamlined CLI: 975 tokens
- 94% reduction
Anthropic's engineering team reported 98.7% token savings switching from MCP calls to code execution.
Vercel recently launched Agent Browser, a Rust CLI for browser automation, when Playwright MCP already exists.
They wanted simpler context, fewer tokens, better outputs.
If you want to start doing the same, here's all you have to do:
1. Find a CLI on GitHub (or have Claude Code build one)
2. Wrap it in a skill with your workflow baked in
3. One command, consistent results
MCPs for exploration. CLIs for execution.
Currently watching the Money Models course after reading the book. Absolutely brilliant. This should be a must watch AND read if you’re doing business in any form.
I didn't expect product research to be this enjoyable
Out of all the tools I looked into, EverBee seems to be the best for Etsy product research
I've got a few ideas for what I want to list, so the next step is to figure out how to create them
Today I started doing more research into products to sell. Also looked into some different AI Image Generation tools like Midjourney to get a better feel for the landscape.
Seems like Midjourney + Etsy has a ton of potential, so I'm going to go that route.
If there is one thing I know for sure is that, for better or worse, AI is here.
You're can either learn to ride this massive AI tidal wave, or be on shore waiting for it to hit you.
Breaking 🚨
None of the OpenAI services including ChatGPT, GPT-3 playground and API is serviceable at the moment.
1000s of apps built on top of OpenAI API are down.
High time to avoid reliance on a single platform, big opportunity for companies like Cohere, Anthropic & OneAI
5. Web Developer Checklist
A checklist of tasks to consider when building and launching a website, covering areas such as performance, security, SEO, and more.
🔗 https://t.co/MfAD1pGoPc
Most people suck at building relationships online.
It's always:
- Help me!
- Can I pick your brain?
- Will you like my Tweet?
A terrible approach.
Here's how the top accounts on Twitter build real relationships:
🧵