What Does a COO Actually Do with AI?

People often ask me what a COO actually does with AI. I don't choose the tools. My job is to find where our team's time and expertise are going to waste, and where AI can create the most value for our customers.

Configuration takes up 35–40 percent of our project time. Around it is a lot of manual work that repeats from project to project: meeting notes, meeting prep, specification spreadsheets and training materials. And because we're human, the quality isn't always the same.

AI Doesn’t Do Our Work. It Makes Room for It.

Right now, AI helps us in three ways.

• More time for the customer's business. When repetitive work takes a fraction of the time it used to, we can spend more of our time on what customers hire us for: understanding their business and challenging it.
• More consistent quality. We've had shared templates for meeting notes, kickoffs and project wrap-ups for a long time. What's new is that AI puts the material together faster and brings in everything we've already agreed with the customer, from meetings and emails to CRM notes. Nothing gets lost when an account moves between teams, and nothing gets forgotten.

• A better starting point for configuration. AI pulls together what we've learned in workshops, emails and conversations and drafts a configuration plan from it. When the details don't rely on one person's memory, there's less room for error. The consultant then reviews the plan, makes the final calls and builds the configuration based on it.

AI doesn’t belong everywhere. Understanding the customer’s business, asking the right questions, digging into their processes and challenging the customer are our job and our strength. So is the final judgment.

Ease Shouldn’t Replace Thinking

The ease of AI is also its biggest trap. When content is effortless to create, it's easy to create too much of it. You stop noticing the volume, and it gets sent on as is. We've all received an AI bomb: ten pages of text where it takes an hour to find the one point that matters. So our rule is simple: AI writes the first draft, and a person reviews it, cuts it down and refines it until it's good. Every time.

A Good Experiment Isn’t a New Way of Working Yet

When one consultant's good idea becomes the way the whole team works, rollout is what matters most. Every new practice is planned, tested on real projects and then put into use. After that, we train the team and keep reinforcing it until it becomes second nature.

Just as important is knowing when to stop. Not every experiment works, and a bad idea doesn't deserve more time just because it has already taken some.

What’s Next?

HubSpot's Breeze Assistant has come a long way, and we use it every day for things like preparing for meetings and pulling together CRM data. It can't yet handle configuration at the scale of our projects, but I'm looking forward to when it can.

Automating configuration is on our roadmap, and we're getting there piece by piece as the tools improve. Until then, our consultants do the configuration themselves. We're also exploring whether AI could check the quality of a portal's setup and flag the right things at the right stage of a project. All of it follows the same path: test first, then roll out.

What’s in It for the Customer?

Customers see this in three ways. Their consultant spends more time on their business and less on paperwork. Quality doesn’t depend on who’s running the project or how busy they are. And because information carries over from meetings to the CRM and from one team to the next, customers don’t have to tell us the same things twice.

I'm not aiming for a team that does more, faster. I'm aiming for a team that does its best work every time, and has the time to do it.

Lotta Laukkanen
COO, Kaksio Labs