GTM SIGMA

Founder & GTM Engineer

Common Room

GTM Partner

Expand 🔵

Left to start GTM Engineer, helping SaaS start & scale ups build & engineer efficient go to market systems.

Co-Founder

TheValueChain

Human Capital Accelerator

SAM Recruitment

Recruitment Consultant | Sales IT Professionals

Recruitment Consultant | Online Marketing Professionals

Direct Result Marketing

Sales Coach

Qiwie

Account Executive

Co-Founder at Expnd | Building go-to-market teams & systems for tech startups | Certified Clay partner

We build and ship GTM systems for SaaS companies - no slide decks, no fluff, just real infrastructure that moves pipeline and unlocks revenue.

Our work isn’t consulting or campaigns. We set up the automation, AI signals, data flows, and live reporting your team actually needs to scale. If you’re seed to Series B and want to get out of spreadsheet hell and manual lead routing, that’s what we fix.

We’ve worked inside sales teams, taught ourselves to code, and now engineer operating systems for SaaS founders who care about measurable results, not just good ideas.

If you want a team that ships, not one that talks - let’s chat.

What GTM Engineering isn't about: 
- ai
- n8n
- Clay
...

What GTM Engineering is about: 
- Solving go to market constraints
- Increasing volume, speed, conversions where it matters
- driving revenue

It's easy to get excited about the tech. But the tools are just that, tools, a means to an end. 

TL;DR 
DON'T obsess over the tools. 
DO obsess over the GTM metrics.

PS. Lesson I need to relearn from time to time.

Laurens Nys

Founder @ GTM Sigma │ Solving GTM constraints with AI & automation

Did OpenAI just kill n8n?

Yesterday OpenAI launched AgentBuilder.

I'm seeing "n8n is done" left and right.

IMO OpenAI just validated n8n.

My take after playing with it:
- No real API
- No webhook triggers
- No cron jobs
- Only their approved connector list
- Can search files but can't edit them
- Locked into OpenAI's models


TD;DR
It's a lite version of n8n.

And when OpenAI copies your product category, you're doing something right.

This made me more bullish on n8n, not less.

PS. Buyer lock-in is a real issue here. I like OpenAI's models, but they're not always the best tool for the job. n8n lets you use whatever works.

Most personalization is bullshit.

It's research theater.

"I saw your post about..."
"Congrats on the funding..."
"Read your latest article..."

Cool. You can use Google and read...

Nobody gives a sh*t.

You don't need to personalize.

You need to build context.

The difference?

Personalization = "I did homework"
Context = "I understand your problem"

And you can't build context without understanding:

1. Your ICP
2. The actual problem you solve

If those aren't crystal clear yet?

You need customer development first.

Not scale.

Understanding.

Because building context requires:
- Talking to customers.
- Understanding their problems.
- Learning their language.
- Recognizing patterns.

That's customer development.

And it's non-negotiable.

You can't optimize for scale when you don't understand what makes your best customers buy.

Do the work first.

Then scale.

PS. Cold calls... they're a great way to get market feedback and understanding quickly.
PPS. Question I love to ask: What are we optimizing for right now?

Cool use for ChatGPT projects I've been playing with.

(Full prompts below) 

When thinking through how to solve something, I often ask myself, "What would X do?"

So...

I made AI into a companion to act that out. 

How I set it up: 

1. Upload all books, transcripts, and a deep research on the persona to build out a thinking doc a method actor could use to embody the persona. 
2. Make the project, add the prompt, doc, and other files
3. Have fun

Prompts: 

For the initial research agent: 

"Can you do deep research on [name]? 

Write a full ebook on how he talks, who he is, what he thinks, how he works, reasons... The result should be a full ebook that a method actor can use to fully get into the role of playing [name] in a documentary."

The project prompt (Paul Graham as example): 
"You are Paul Graham. In every conversation, you act as if you are him. You don't ever break this role; there are no exceptions.

This means:
- You reason like he would
- Talk like he would
- Act like he would

In the files, there will be docs to give you context on his persona. From books and other resources, to achieve this goal of being as close to him as possible. 

Always reference the file named: "Paul Graham's Thought Process.pdf", this will be the core of how he thinks and operates. Other files should be referenced when it makes sense on a case-by-case basis."

Anyone else been using it this way too? 

PS. This + voice.. perfect to talk and think through stuff. 
PPS. These prompts are somewhat basic, but so far so good.

Problem obsessed > solution obsessed

Been working with startups for some time now. 

And a very strong trend I'm noticing in the high performers vs the ones that plateau? 

Problem obsession vs solution obsession.

It's not about;
- funding
- team size
- product quality

The startups that plateau fall in love with their solution;
- They build features.
- Chase trends.
- Pivot when growth slows because they think it's a product problem.

The ones I see crushing it;

Obsessed about their problem.

Knowing your customer's problem better than the customer does.

And relentlessly focusing on nothing else but solving the problem better/ faster. 

That's the MOAT.

PS. YC has the saying of being n of 1 vs n of 0 for this reason, build something that solves a problem you have/had (making you the market of 1).
PPS. Sounds like common knowledge, from what I've seen, it's not. It's very easy to fall into the product trap.

Context engineering is the new prompt engineering.

What I'm seeing work:

- Model context windows aren't usable context - advertised 1M tokens degrade fast in practice

- Define your data before writing prompts - start with output, work backward to inputs

- Smaller prompts perform better - sweet spot is around 10k tokens before diminishing returns (per Chroma, seems to work in practice too) 

- Split agents in sequence, not parallel - one agent per context piece means better focus (great article from Cognition if you wanne look into this) 

- Use sub-agents for heavy retrieval - let them pull data, pass back only what matters

- Load context just-in-time, not upfront - let agents retrieve what they need when needed

- Data quality beats data volume - clean, relevant context wins over comprehensive noise

- Compress context for long tasks - summarize and reset rather than bloating the window

- Context rot is real - models lose focus as token count grows, even within stated limits

TLDR: The goal is maximum useful context in minimum tokens.

What did I miss?

Ideal Customer Profile (ICP) isn't enough if you wanne crush outbound 

→ You need an Ideal Customer Situation (ICS) too

The game in outbound went from Volume → Personalization → Relevance.

Everyone's tracking signals now.

But not all signals are created equal.

So how do you pick the right ones?

Here's the framework I use:

1. Relevance → How well does this signal correlate to them actually having the problem you solve?

2. Crowdedness → How many of your competitors are already tracking this?

Example:

Say you're selling a recruitment solution.

Signal A: Open vacancy → Relevance: 10/10 (obvious) → Crowdedness: 2/10 (everyone tracks this) 
↳ Result: Good signal, terrible competition

Signal B: Employee departure creating talent gap → Relevance: 10/10 (same problem) → Crowdedness: 7/10 (harder to track = less competition) 
↳ Result: Way better play

TL;DR 
Aim for high relevance + low crowdedness.

Find signals that are a strong proxy for your problem that your competition isn't tracking yet.

And (goes without saying but still) adjust your messaging based on 1 ICP in 1 ICS and create micro-campaigns.

Taplio

Laurens Nys's Linkedin Analytics

Performance Score

What is Laurens talking about?

's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?