What happens when your customers can do more themselves?

I have built my Wingman using Claude Code and ChatGPT Codex. It helps me research customers, manage my calendar and work out the most valuable thing I can do next towards the goals I am working on.

When I need something to work differently, I can describe what I want and ask it to investigate, build, implement and operate it with me. That sounds technical. In practice, I explain what I want in ordinary language, observe what happens and make adjustments.

It has made me think about the companies I buy from. As I become able to do more myself, what will I still need their help with? And what happens to my own business when my customers gain the same capabilities?

DHH recently wrote about how much software there is still to build as development becomes cheaper. What interests me is what that makes possible for the person using it. More of us can do something about the tasks we have accepted as cumbersome, expensive or dependent on someone else's availability.

That changes the decisions we make about buying help.

A business owner might previously have paid someone to gather information, compare options and prepare a recommendation. With AI, they can do more of that preparation themselves. They can arrive at a conversation with a clearer understanding of the problem and several possible ways to solve it.

They may still need help. But what they need help with has changed.

If I have already used my Wingman to research a company and prepare for a meeting, I want the conversation to move beyond information I already have. I want the other person to help me understand something I have missed, challenge an assumption or work through a decision I am unsure about.

The preparation makes me able to ask more of the conversation. It can also make me aware of how much I still don't understand.

I know this from the way I work with my own Wingman. It knows my goals, my strengths and weaknesses, and the parts of running a business that drain my energy. That context helps it take on work that I would otherwise postpone and helps me spend more of my attention on valuable decisions.

I still have to check what happens on my behalf. I am the sender, and I carry the responsibility. Having more capability available to me means I can act on more things, which also means I have to be deliberate about what I choose to do.

I think something similar will happen for customers. Being able to produce a plan or investigate an opportunity gives them more options. It also leaves them with decisions about which options deserve their time and money.

For those of us who advise companies, that is worth paying attention to. A customer who has done their own research may need us to examine the reasoning, understand the constraints and help them make a choice they can carry through. We need to understand what they can already do before deciding how we can help.

Andrew Chen explores a related issue in his analysis of AI agent businesses. Access to an intelligent agent does not automatically give a company a lasting advantage. He looks at how relationships, shared context and reputation could make a service more useful.

That resonates with how I have built my Wingman. Its usefulness depends on knowing enough about me and my work to help in the situation I am actually in. A customer relationship needs that kind of understanding too. We have to know what the customer is trying to achieve and where they are having difficulty, including how their use of AI is changing both.

I want to use AI to do more of the work around my decisions. I expect my customers to want that as well. When we work together, I want to understand what they have already figured out, where they still need help and what we can get done together that they would struggle to get done on their own.