From chat to agents: how my mostly female team rebuilt our business in two quarters

Our companies work in life science and genomics. My team is mostly women, in science, sales, operations and administration. A year ago, very few of us would have called ourselves technical.

After a slow start with AI early in 2026, and an expensive bootcamp in the middle of the year, something finally changed between the second and third quarters. We stopped using AI only to chat, and started using it to build.

Stage one: chat tools

Like most people, we began with simple chat tools. You type a question and you get an answer. They helped with emails and ideas, but every answer started from zero. The tool did not know our business, our customers or the way we like to work.

It was useful. It did not change how the business ran.

Stage two: building with an AI coding assistant

The turning point was an AI coding assistant. Instead of asking it questions, we described what we needed in plain words, and it built it: the pages, the forms and the connections between them. We read the result, tested it, and asked for changes until it was right.

Our first big project was the company website. We rebuilt the whole site this way. Then we went further.

We built portals: one place where our team signs in to reach the work they need, and portals where our clients sign in to see what is theirs. Things that used to sit in emails, spreadsheets and people’s memories now have a proper place.

Stage three: helpers that work on a schedule

The newest part, and the part growing fastest, is our routine AI helpers. These are small AI agents that each do one routine job on a schedule: every morning, every week, or whenever something new arrives.

For example, when someone sends an enquiry through our website, a helper prepares a first draft of the reply. A person on the team reads it, changes it and decides whether to send it. Other helpers prepare summaries and checks, so problems are seen early instead of late.

The rule never changes: the helper prepares, a person decides. No message written by a helper goes to a client until one of us has read it.

What changed for the team

The biggest change was not the technology. It was the role of the people.

Women on my team who never thought of themselves as technical now describe the tools they need and see them built. The skill that matters most turned out to be clear thinking: knowing what you need, explaining it simply, and checking the result with care. Many women already use those skills every day, at work and at home.

Routine work that used to fill whole afternoons is now prepared for us. That leaves more time for the things that need a person: talking with clients, solving real problems, and looking after each other.

We also became more careful, not less. When a helper can act quickly, you think harder about what it should and should not do. We write down rules, we check results, and we keep private information private.

What I would tell another woman leader

• Start small, with real work. One task, one owner, one simple measure. I wrote about this in my journal entry on our bootcamp.
• Let the people who do the work describe what they need. They know the work best.
• Keep a person in charge of every decision. The AI drafts, prepares and checks. People decide.
• Expect it to feel slow, then quick. Our second quarter felt slow. By the third, the team was building its own tools.

Why I am sharing this

Two quarters ago, I would not have believed that my team could build its own website and portals. We are not a technology company. We are a life-science business with a team of mostly women who were willing to learn.

If we can do it, many women can. That is why I started Myself + AI-Self: to share what we learned, the good parts and the costly parts, so that other women can begin with more confidence and less waste.