How to Use AI for Competitor Research and Cut Manual Work

Many people think AI is just a toy. They use it to generate funny images or write obvious, template-heavy social media captions.

But in the hands of people who know how to use it, this tool cuts work that used to take days of manual effort.

Batch 1 of AI Circle was attended by 23 people across roles. PMs, marketers, a support lead, and ops. The point I wanted to prove there was simple: AI is not just a toy for tech teams.

This piece is about the method, not a promise of results.

An Old Habit of Finding Efficiency

Looking for efficiency gaps is an old habit of mine, long before AI became a trend.

Across my years leading products at Tokopedia, Hijra, and Flip, my work always came back to the same question: which part of this process does not need to exist. Process improvements, cost optimization, and product decisions that simply waste less.

The difference between then and now is the tooling.

Back then, creating that kind of impact required a senior position, an engineering team, and systems that took months to build. The Tokopedia and Gojek account integration is one example, coordinated across teams and across companies.

Not anymore. The 8-channel content distribution system running for my own brand is something I built and operate by myself, with no agency team. What changed is not the person, it is how far one person can now reach.

Using AI for Competitor Research Without the Grind

Take the task people avoid the most: competitor research.

The old way is opening a dozen tabs, copying competitor data into a spreadsheet one by one, and only then starting to think. The thinking gets whatever time is left, because half the day went into collecting.

What I teach is not one magic prompt. It is three turns.

In the first turn, the AI only reads and summarizes the public documents you give it. Annual reports, pricing pages, product changelogs, anything genuinely public. It is not allowed to conclude anything yet.

In the second turn, you ask it to compare: who offers what, at what price, for whom. The output here is a table, not prose.

In the third turn, you ask where the gaps are. Which part nobody is covering, and why that makes sense.

Same task, different order, very different output. What makes the result sharp is not prompt length. It is you knowing which question comes first.

Why a One Day Workshop Can Change How You Work

People often ask how a single day can make a difference.

The answer: we do not study theory.

Most AI classes out there teach prompt lists or the newest tools. The problem is that tools change every week. What you learn today can be stale next month.

In AI Circle, the focus is solving real work problems. We learn to define the problem first, map where the workflow leaks time, and only then pick the tool that cuts that specific part.

In class, I teach how to map one process end to end. Break the process into small steps, separate the human part from the AI part, and only then build the system.

Having managed 4 product teams at once, I know how hard cross-team coordination gets. Small initiatives that only need data cleaned up sit in the engineering backlog for months because they lose to bigger priorities.

Once non-technical people can do it themselves, that queue disappears.

This is not about cutting headcount. It is about freeing people from monotonous tasks so their time goes into thinking.

Learn the Method Directly

AI Circle Batch 2 runs on Sunday, 9 August 2026. The format is hands-on rather than a demo: you bring your own work, and we take it apart together until it runs.

Registration is on the AI Circle page.

If you have a team you want trained directly with a curriculum shaped around your own internal problems, see the Corporate Training page.