How Curiosity Leads to HAL

Every revolution starts with a question. For the last two years, a single question has been bouncing around my head:

What if we could build an AI system which acts like a team member, fully autonomous, adaptable and self learning, so it can perform commercial work for us (as we sit back and hold our breath)?

That's how our idea for HAL was born, as a real moment of curiosity, skepticism, and (I’ll admit) a bit of mischief. I have always loved experimenting, especially those that break the status quo. This particular experiment was not just to chase efficiency, it's a genuine question about how far AI can become a team member. I wanted to see how far we could push the boundaries of AI—and, yes, have some fun along the way.

01 | Moving Beyond the Norm

Anyone who knows me knows I’m obsessed with commercial perfection and efficiency. My natural tendency is to standardise approaches, capture it in a process and run ruthless data analyses to iterate. Over my years in sales I have learned that perfection isn’t about rigid processes or endless spreadsheets. It’s about staying flexible and finding simple ways to solve problems—ways that are simple, effective, and, if possible, a little bit mind bending.

In reality both are true- commercial operations are full of repetitive, time-consuming tasks: prospecting, outreach, research, data entry. I’ve spent enough hours in the trenches to know how much time and energy these eat up. So I started asking myself:

Isn't there a smarter way? What if an automated solution is smart enough to adapt yet simple enough to execute processes?

That's when the idea became real. Not just to automate the boring stuff (and get the AI emails everyone so quickly started getting and ignoring), but to reimagine what’s possible when you let AI take the wheel. Can we create a system that does not just follow instructions, but makes decisions, learns and rewrites its own instructions, all while delivering real world impact?

With this mindset we set out to create HAL, to challenge our perspective of what AI based systems can do and be.

02 | Why HAL? A Name With a Wink

Every experiment needs a name. After a few rounds of debate (and ruling out the many options that were boring), we landed on HAL. Officially, it stands for “Hyper Adaptive Loop,” reflecting the system’s core design: a continuous cycle of action, feedback, and improvement.

But, truth be told, I also liked the idea of the “Hyper Autonomous Lord”—a tongue-in-cheek nod to just how much control we were willing to hand over to our digital creation. Plus as an avid fan of '2001: A Space Odyssey', a reference to HAL 9000 was just too good to pass up. (Don’t worry, ours is programmed to be a lot less ominous.)

Choosing the name was just the first step—next came the real work.

Choosing the name was just the first step - next came the real work.

03 | The Architect’s Table: Sketches, Data Flows, and Self-Learning Dreams

The reality of HAL was not a clinical well organised machine, it was rather a chaotic workshop full of ideas. The inception consisted of late-night sessions at my desk, with half-finished architecture sketches and data flow diagrams. I was trying to solve a complex question—how can HAL be a self-learning adaptive system, whilst adhering to our continuously updated best practices and methods (guardrails), without locking it into just following instructions?

I’d draw out one approach, only to realise halfway through that it missed something essential. In some cases sketches became too complicated and start over from scratch. It was chaotic, frustrating, and (honestly) a lot of fun.

Building HAL's components was an adventure in itself. We built (and are building) one agent at a time. Each of those relies on another, and every one of them plugs into HAL's central knowledge. Getting them to work together is (this is a suggestion AI made on how I should describe the process) 'like herding cats'. That's rather accurate, we change something in one agent and see others adapt, not always in a way we expect or want. It's an awesome journey of experimentation and hypotheses.

04 | The Real Premise: Scaling Our Proven Practice

The premise and job for HAL is not just to do sales—its goal is what truly sets it apart. HAL autonomously automates simple sales operations, but does that as if it were one of our team. Its foundation is Gysho’s (and my own) fact-based, proven sales and project methodology. Over the years, we’ve developed a way of working that delivers results—structured, data-driven, and relentlessly practical.

The real goal of HAL is to take everything we’ve learned, everything that makes Gysho effective, and bring it to the world at scale. Automation is just the mechanism; the real magic is embedding our best practices into an AI that can apply them consistently, tirelessly, and—most importantly—beyond the limits of our own team. If we can get HAL to work as we would, and show real life impact and outcomes, we can do that for so many more cases.

Sales is the first use case, because that’s where our need was most urgent. The vision and real application is much bigger: HAL needs to operate like a team member to apply our expertise at a much wider scale. Its that end goal which keeps us pushing HAL further.

05 | Wrestling with Big Questions

As HAL started to take shape, I ran into questions that went far beyond code and architecture:

How can we give AI enough knowledge and logic to trust it with real world decisions?
Can we manage transparency, governance an accountability in a fully autonomous self learning system?
Can AI-driven processes be not just efficient, but also ethical and act like a (rational) human?
How much can we truly leave to an AI system, before it just becomes 'too much'?

These questions are not just academic debates, they're real problems of ethics and acceptance of AI in daily operations. They shaped every design decision, from how HAL interacts with customers to how it reports its actions. I didn’t want HAL to be a black box—I want it to be an extension of us. Not a ChatGPT clone, but something that truly delivers value to whoever interacts with it (or is it a he or she, we are yet to find out).

And as we keep building, these questions only get more interesting.

06 | Feeding Our Team's Inner Inventor

The most rewarding part of this ongoing experiment is not HAL itself, it’s the fact that it continues our internal drive to experiment and push boundaries. Gysho started out as a big experiment, building things that (at that point) no one dared ask for because it 'can't be done'. Whilst HAL does not come cheap (tokens, systems, R&D time), it feeds our real need to build in things that break the status quo.

As we were building the first components of HAL, it’s pushed that beyond expectations. We ran into a myriad of questions and problems that challenge our thinking and deliver new perspectives, which in turn are useful in our client side projects. I view it as an investment, where we make sure we don't get stuck in the status quo of what others say is possible, we create the new norm and as a result create better solutions to real world problems.

The spirit of invention is what makes every late night and every challenge worth it.

What is Next? | Join the Journey: Exploring, Building, and Shaping HAL Together

This blog marks the beginning of HAL—a bold experiment to create an AI system that acts as a true team member: autonomous, adaptive, and always learning. Driven by curiosity and a desire to break the status quo, we share the story behind HAL’s conception, the challenges of building it, and our vision to scale proven expertise through AI. Join us as we document HAL’s journey from sketches and late-night ideas to real-world impact.

CURIOUS TO SEE HOW HAL EVOLVES?

Follow our journey, share your thoughts, and help us push the boundaries of what AI can do. Your questions and ideas could shape the future of HAL!

JOIN THE CONVERSATION

Tags:

Post by Sander de Hoogh

Enterprise AI Adoption in 2025: Best Practices for Sustainable ROI

3 min read

AI Agents in 2025: Business Impact & Leadership Strategies

4 min read