Why I built pleko

Not a decision to start a consultancy. A name for a shape of work that never had a job title.

The last post ended with a line I had been circling for years: I stopped trying to get through the door and built pleko instead.

I have struggled to define what pleko actually is, to turn it into a short pitch, and to explain it without making it sound like another AI buzzword.

So here it is. pleko is one person who works out what is tangled in your business, builds the fix, makes it safe and teaches you to run it.

It didn't start with a business plan

I never sat down and decided to start a consultancy. If you had asked me three years ago what I wanted to do next, I would have said: find a good role in a company where this combination of things is useful. That is what I had been trying to do for a long time.

What I was good at did not fit in a job title. I could understand a business and its customer well enough to see what was actually broken, and I could usually see which part of the company owned the problem. I could sit with marketing in the morning and operations in the afternoon and notice that they were describing the same thing with different words. I could take a messy process and turn it into something with steps, owners and a result you could check.

None of that is a job title. It is a set of fragments borrowed from several: marketer, analyst, project manager, operations lead, consultant. Companies rarely hire for the combination, and never as one role. The roles I did hold always came with a boundary, and the useful part of my work was almost always on the other side of it.

I had always been drawn to the nerdy stuff: how things work behind the scenes. That curiosity had not always worked in my favour. With pleko, I could finally put it to work.

The last post was about that mismatch. This one is about what happened when I stopped trying to fix it.

Two things changed

The first was practical. The distance between understanding a problem and building something that fixes it got much shorter. I am not a programmer. Ten years ago, if I saw that a process could be automated, my job was to write it up well enough to convince someone with engineering time to build it. Now I can build a working version myself, test it, and find out whether the idea survives contact with the real work. If it does, the engineer gets a much better brief. If it doesn't, nobody has spent three months finding out.

The second was less obvious. I stopped needing a company to give me the container.

That is the part people skip when they talk about AI and work. The tools did not just make me faster at tasks. They made a shape of work viable that had no natural employer: one person who diagnoses a process, builds the fix, makes it safe and teaches the team to run it. Inside a company, that work is always split across three departments and two approval chains, and it usually dies somewhere between them.

What pleko actually is

It started to feel real when I chose the name. pleko comes from the Greek for to weave, to connect. That was the work: building the often invisible infrastructure between old technologies, new ones and the people who use them.

It is not an agency, and it is not software. It is one person who understands the business and can build the thing, working directly with the people who will use it. No account manager, no being passed between people, no platform you have to move onto.

The work has four moves, always in this order (the services page has the details).

Diagnose the process. Before any tool. What happens today, who does it, how long it takes, where it breaks, what it costs when it breaks. Most of the value is here, and most of it is unglamorous: watching how a team actually works rather than how the org chart says it works.

Build it. Usually in the tools the client already pays for. A flow, a database, a model, a form. The goal is that nobody has to learn a new system to get the benefit.

Make it safe. Permissions, data protection, a human checkpoint where it matters, a log of what happened. This is the least interesting part to talk about, and the part that decides whether the thing survives its first real month.

Teach the team. Written guides, a recorded walkthrough, a workshop. The measure is not whether I can run it. It is whether they understand it well enough to run it day to day.

A real example, with the details changed enough that the client stays anonymous. A large company with around thirty brands gets questions from customers and from its own sales team all day: by web form, by email, by message. The people who answer them are brand managers and the customer team. Every answer is written from scratch, in a hurry, between other work. Nobody had ever written it down as a process. It was just what everyone did.

The interesting question was not "can AI answer these questions?" It was: which answers already exist and are approved, who signs off on a new one, what happens when nobody knows, and what the system should refuse to answer. We spent more time on those four questions than on the model.

That is what untangling first means: untangle, then automate. Sometimes the problem is not the tool at all.

The three promises, and the skill behind each

Untangle first. The skill is asking questions in the right order, and being willing to find that the problem is not where the client thinks it is. Many of the projects I am asked about turn out to be a process problem, not a tool problem, and the honest answer is sometimes that a spreadsheet and a weekly meeting would do more than an automation.

It's yours. Your accounts, your data, your documents, your tools. I do not want to build something that holds a client captive, charging them to maintain a system only I understand. If they want help expanding it later, great. But it should be theirs to use, and they should understand it well enough to say what they need. Making that real takes knowing how the pieces fit together and where ownership can quietly get lost.

I'll teach you to do it yourself. The skill is explaining. If I am the only person who understands what I built, I have created a dependency and called it a solution. This is the promise that costs me the most work in the short term, and the one that makes the other two believable.

What is still hard

Explaining the value before the work exists. I was surprised how many people could use this, from freelancers to small companies. Once something works, the time saved and errors avoided are clear. Explaining that beforehand, to someone who may not know this help exists, is still hard.

Saying no. The projects that do not fit teach you the most about what you actually do, and turning them down is taking me longer than I expected. I still sometimes take something on because it is interesting rather than because it is a good fit. The difference is that I notice sooner.

Being one person. Sales, delivery, invoicing, the website, the newsletter. The generalist's advantage is also the small business's problem: there is always another piece to hold.

What I got instead of a job title

Clients do not need to decide whether I am a marketer, an operations person or a technical builder. They bring a problem that crosses those lines, and we work out what to do with it. Nobody has to believe a story about my career. The work either helps or it does not, and that is settled in weeks rather than in an interview.

I would like to tell you this was a plan. It was closer to a series of experiments that kept working, until the shape became clear enough to name. pleko is that name.

In the next post, I want to look at what happens before the tools: how to spot a tangled process and find the problem worth solving. Sometimes we don't even notice a process is tangled: we've lived with it for so long that we assume that's just how things have to be.

If you have spent years being the person who connects things rather than the person who owns one thing, I would like to hear what you did with it. Write to me and tell me.

← All posts