Escalis.ai · Case Study

A deal closer with AI expertise, a computer that stays on, always a human in the loop

What I set up for clients, I use myself. The Claude case study on this site is how I wrote, built, and delivered Not Yet. This page is the next chapter: how a full Grok Bot team runs the commercial and ops loop, so the hunt, the writing, and the follow-up keep moving while I am on the road.

Seven-League Boots leap from Today status quo into With Grok Bot
This is how Grok Bot feels to me: like the Seven-League Boots fairy tale I read as a kid. Each step covers huge ground. I put the boots on, and I am empowered to keep the helicopter view while the daily commercial and ops loop still gets done.

I am not an AI model designer. I studied artificial intelligence at KU Leuven in 1990, as part of my mechatronics curriculum. It was interesting but unusable. No fast processors, no cloud, no storage existed at that time. And more recently, on a beach in Pattaya, I picked it up again. I took the hard Stanford lectures so I would understand the ground. Then I chose not to go and train AI models. That is work for people who want that life. Mine is sales, business development, and getting a real result for a company. So I learned the tools that a one-man operation can actually run: Claude Cowork and Claude Code for my Not Yet memoir and the websites NotYetMemoir.com and Escalis.ai, and Grok Bot for the daily work that has to happen whether I am at the laptop or not.

One rule, same as the other case study: the final word is mine. Nothing goes out that I have not read.

Why Grok Bot, not another chat window

Claude Cowork is excellent when I am in the files: episodes, photos, a sound-like-Frank pass. I hit a different wall with the commercial work. Finding new prospects, job hunts, intro and cover letters, a CRM, reader research, posts, inbox signal, costs, travel. That is not one conversation. That is a small operation. I wanted something that keeps working when the MacBook lid is closed, talks to me in the morning, and files the work where I can approve it.

I found Grok Bot on X. X is my news platform. It is also where I found Anthropic, in Pattaya, which became Episode 27, a free read episode on NotYetMemoir.com. So this is not a random new toy. It sits on the network I already live in.

What the team does now

I talk to one contact: the Chief of Staff. That bot coordinates the others and delivers a morning brief around 8am Phnom Penh time, or morning wherever I am, with the actions I need to take: social posts, email follow-ups, job applications to review and send, and whatever else is due. Behind that sit specialists. Each owns a job. They hand work to each other when it helps. I invest significant time training each bot until it does the job right. I set the objective. I stay on send.

The VC lane stays light in public: the bot builds the map, I do the outreach, and the parachute story is the hook. Details when someone asks.

Pitches still come from my real CV, 25+ years. VaultSpeed, AiVidens, FINN Banking of Things (ING), SettleMint, OTN Systems (Belden), Transmode Systems (Infinera). The $400M Spike deal. The Transmode channel. Human in the loop on every letter.

Two brands, my hand on send

Escalis.ai generates the cash: consulting, sales, AI implementation for companies. LinkedIn stays here, a few times a week.

Not Yet is the long book. It has its own X account. Book posts go out on X as @NotYetMemoir, never as @EscalisAI. Same separation on Instagram, Facebook, Pinterest, and Patreon Quips. The bots can prepare the text and the picture. I still decide. Nothing is applied, posted, or emailed without me.

The photos are mine. Thousands from the road, a few hundred hero shots. Watermarks come from my own developed Mac tool, diagonal notyetmemoir.com, copyright 2026 at the bottom. Titles can sit on those pictures. Faces and number plates can be blurred.

How the work is actually split

Grok Bot works the way I used to run a sales organisation. A team of AI agents perform work, talk to each other, and hand work across, all toward the objective I set. Chief of Staff is the person I talk to, but I also talk directly to the specialist bots. They take the jobs and report back.

When I ask for a job hunt, one worker searches and verifies postings while another writes pitches from my real CV and files them as cards on Trello with the right color tag so I know instantly what type of job it is (executive sales/BD, versus AI consultancy only, or a VC contact). When I update my own biography, a worker goes back over every card and every letter and rewrites them. The morning brief is waiting when I sit down, with the actions for that day lined up.

Each specialist bot took real training time before I trusted it with the job and the Chief of Staff helps and coordinates. Social packs, reader research, inbox signal, costs, and travel run as their own jobs while I work on something else. I never have to hand work from one to the other. That handover happens without me.

And the real nice thing is that I often work directly together with one of the bots because I want to train them properly on the tasks they perform, and it never upsets my Chief of Staff which would not be the case in the real world. As a matter of fact, the specialist bot will inform the Chief of Staff bot proactively about our dealings. I really have to smile every time this happens.

And my marcom and posts in the right format and style, I all run this by myself. No need to ask for marketing budget and go in the queue like in the real world, and getting awkward feedback that there is no budget and time for this. The entire thing is usually done within the same minute, and it puts every time a broad smile on my face. Sending out the post, Done, Next!

What this is not

It is not unsupervised posting, and it is not unsupervised job applications. Approval sits with me.

I do not train large language models. I use them. I sell with them. I will tell a client the same honest line I tell myself.

The Grok Bot team owns the daily commercial and ops loop: prospect and job search for Escalis.ai, the VC parachute lane, readers for NotYetMemoir.com, social packs, inbox signal, costs, travel, and the rest of the list above.

And how does this feel?

When people ask me how it feels to work in my new setup that I will never abandon again, I will only keep on improving it, I tell them that it feels like being in the fairy tale that I read when I was still a young kid, and it is called the Seven-League Boots. It is about a small boy that stole the boots of a giant, and put them on and with each step he could span seven leagues. One league (roughly 3 miles / 4.8 km) was considered to represent the distance walked in an hour by an average man. So seven leagues would be about 21 miles / 34 km, or equivalent to walking for 7 hours. It really feels like this. I can perform work now a hundred times faster and a lot more efficient, and not feeling overstressed or whatever else. In fact, I can keep my mind focused on the helicopter view while executing the daily grind with my Seven-League Boots.

From today's status quo into With Grok Bot. Seven-League Boots, in motion.

In the past my calendar had blocks of two or three hours with no meetings, just so I could catch up on email, write a report, or prepare a client proposal. My whole work life was serial processing, like the old RS-232 interface: I delivered the information and the work myself, one thing after another. A parallel printer interface like LPT was a step up, but that analogy is not strong enough for what I run now. Today I have many parallel lanes at once, and the work in each lane is done by the bots. I feed a bit of input, I steer, I say do this, do that. Then the agents execute. When a bot is trained, it becomes more self-sufficient. It talks to other bots. It updates the Chief of Staff. I get the morning brief and updates during the day. This keeps going while I sleep, while I have lunch, while the laptop is closed. That is the breakthrough: not only parallel process, but parallel work performed by agents at quality that holds up. I still keep the last word. I read what goes out. I correct if needed. Then I hit send.

Why this matters for your business

Most owner-led companies do not need a lab. They need a hunter who still understands the product, a way to keep work moving when the founder is traveling, and a hard stop to check what is going out. A full specialist team under one Chief of Staff is how I run that for myself. The same pattern can sit inside a company that wants the same loop without a department of chat windows and a lot of time consuming meetings.

That is the setup. Grok Bot for the daily commercial and ops loop. Me in the middle. If you want a closer talk about how this actually works, not a slide about agents, let us talk. The memoir, the Storymap, the sites, and the image toolkit are in the Not Yet case study. If you would like to know the person behind it, here is my one-page CV.

Want this for your company?

If you want to see what practical AI could do in your business, let us talk. No jargon, no oversell.

Book a free call