Arun Anderson builds practical agentic AI.

I’m Head of Agentic AI at Quadient and the creator of SmartFlow. I work where enterprise problems and practical AI meet.

This site is the personal side of that work: small experiments, useful explanations, and a place to share what I’m learning.

Find me on LinkedIn (opens a new tab)

HOW IT ALL STARTED

411 attempts, then the app was blocked.

It all started with one pre-loaded app on my Android phone, quietly sending data somewhere I couldn’t see. I wanted to block it. What followed was 411 attempts at flashing recoveries — TWRP, CWM, brute-force combinations, bricking and recovering the phone, then trying again. On the 411th attempt, I finally got through. The app was blocked.

THE PATTERN

The pattern continued

The next time, I wanted to download data from a public website. It was spread across 40 pages, and every page meant clicking through, copying and pasting several sections by hand. I wanted to scrape the entire site in one go, so I built my first scraper. More than 300 attempts later, it finally worked.

The pattern was the same. I don’t confuse the first failed approach with a failed idea.

THE SPACE BETWEEN

Between customers and code.

The professional version of me looked different on paper — technical support, project management, customer success. But it was the same pattern wearing a lanyard.

Every role put me in the space between. Customers on one side, describing what they needed, usually in the language of frustration. Engineers on the other, asking what exactly to build. I became the translator: I'd sit with the problem until I could say it plainly, then carry it across as something a team could ship.

For years, that was the job. Seeing the thing clearly — and handing the building to someone else.

THE TRANSLATOR BUILDS

When the translator started building.

Then AI changed one thing, and it changed everything else. The translator's skill — a clear task, the right context, honest checks on the result — turns out to be exactly what AI needs to build well. The brief I once wrote for an engineering team can be a working prototype by evening.

I still sit in the space between. The difference is that now, when I see the thing, I build the thing — and I leave the method visible so anyone can follow it. That's what this lab is.

How I approach the work

Can someone actually use this? What happens when it’s wrong? Is a bigger model worth it? What should still need a human decision?

Those are the questions I want this lab to make easier to explore. Some things will be polished. Others will be experiments. The difference should always be clear.

I use AI to help build. I still own the decisions about what gets shared.

Role, recognition and learning

Today
Head of Agentic AIQuadient · Creator of SmartFlow
2026
Applied Agentic AI for Organizational TransformationMIT Professional Education · Professional programme certificate, July 2026
2026
AI-Inspiring Innovators for Digital — GoldQuadient · April 2026
2025
Top 100 Customer Success StrategistSuccessHACKER · October 2025

Credentials and role are based on my published LinkedIn profile. This is a personal website, not an employer or education-provider endorsement.

Talk about a practical AI problem

I’m interested in thoughtful conversations about building useful AI, sharing lessons and making the technology easier to understand.

Read the professional overview

Start a conversation on LinkedIn (opens a new tab)