If You're Lazy Like Me, the AI Era Is Doing You a Favour

Let me start with something honest: part of why I got into automation was laziness.

Not the bad kind — the kind that looks at a repetitive task and thinks, wallah, there has to be a better way to do this. The kind that gets annoyed watching good people do work that shouldn't need people.

Last year I found out — after a late diagnosis — that I'm a neurodivergent person. Yup. That might be a whole other post, inshAllah. 😄 But it explained a lot about the way I've always worked: the obsession with systems, the low tolerance for doing the same thing twice, the deep focus when something actually interests me, and the complete shutdown when it doesn't. Turns out "lazy" and "neurodivergent" have more overlap than people think — and automation was always my workaround for the gap between what my brain resists and what actually needs to get done.

Part of it was also geekiness — I grew up taking things apart to see how they worked, running home labs, building small systems just to see if I could. And part of it was what people around me called perfectionism, though I've managed to lower that a bit over the years. AlhamduliLlah for growth. 😄

The core idea was always the same: systems don't get tired, don't forget steps, don't have bad days, (beside bugs). If something can be systematised, it should be.

Then agentic AI arrived — and honestly, it felt like someone handed me a playground full of toys.

The project

A few weeks ago I was supporting a project where the sales team was doing what every sales team does.

Manually searching for companies that might be a good fit. Digging up the right contact. Verifying emails. Writing outreach one by one. Logging it somewhere. Following up when they remembered. Good people doing work that a well-built system could handle better, faster, and without needing a coffee break.

So we built one.

It starts by searching — leads that match a specific profile, tenders in a particular industry, potential partners in a target geography. Not when you ask it to. On its own, on a schedule, while everyone's asleep.

What it finds, it researches. It digs up the right person, verifies the contact, builds enough context about the company and situation so the next step isn't embarrassing. This part gets the most attention, because the quality of everything after it depends on the quality of this.

What it researches, it writes to. Specific, short, sequenced — first touch, follow-up, the whole cadence — drafted and ready based on who the contact is and what the context tells us about them.

And then it either waits for a human to approve before anything goes out, or it sends autonomously. Your call, depending on how much you trust it that day. Both are valid. 😄

The goal: nobody on the team doing that work manually. And when the system is running right, that's exactly what happens. They focus on the conversations that actually need a human — the ones where judgment, relationship, and nuance matter. The engine handles everything upstream.

What makes this different from a basic automation

This isn't a few steps connected together in a simple workflow tool. The system reasons — it makes decisions based on context, handles edge cases, and fails loudly when something goes wrong instead of silently breaking and leaving you wondering why nothing came out the other end weeks later.

That last part — reliability — is the unglamorous work that separates something you can actually run from something that looks good in a demo. Every step logs what happened. Every failure notifies a human. Every retry has logic behind it, not just "try again and hope."

I won't go deep into the full architecture here — for the geeks, the complete diagram is on GitHub (link below). It covers the orchestration, model routing, enrichment cascade, memory layer, and fallback chains in detail. What matters for this post is the outcome: a sales process that used to need daily human attention now runs on its own, and the team's time goes somewhere more valuable.

The bigger idea

My background is a blend that doesn't fit neatly into one box — I've built tech products, ran a coworking space and accelerator for seven years, designed programs for the World Bank, and advised founders across the region. That combination gives me a way of looking at these things that's more holistic than purely technical.

When someone comes to me with an AI automation or product idea, I'm not just thinking about whether we can build it. I'm thinking about whether we should, what it changes for the team, whether the business is ready for it, and what breaks if we get it wrong.

That's the lens I bring to every build. The geekiness and the laziness together, alhamdulillah. 😄

If any of this resonates

If you're building something — a business, a product, a process you want to fix — or even just thinking about it, let's talk. Maybe there's something we can pull together.

It doesn't have to be a big formal thing. Just text me and we take it from there.

Full system architecture on GitHub → [link here]