Your business will need an agent, just as it once needed a website

In 1998 many companies thought having a website was little more than a whim.

For a couple of years, they were probably right.

Then they stopped being right.

And the companies that arrived late paid a price that appeared in no forecast: they didn’t lose a particular campaign or a one-off opportunity. They lost a channel.

I think we are reaching a similar moment, though with one important difference.

This time it isn’t only about having a presence on the internet.

It is about having someone to talk to.

The storefront stops being looked at

A storefront is passive.

It shows what you have and waits for somebody to come in.

For thirty years that worked quite well, because on the other side there was a person willing to look, compare and click.

But we are delegating more and more of that work.

Instead of opening ten pages to compare options, a lot of people now start by asking an assistant directly. And when they get an answer that works for them, they stay there.

They no longer necessarily compare on their own.

They delegate.

Your website can be perfectly designed and still not be part of that conversation.

Not because it is badly made, but because it is meant to be read by a person, not for another system to understand what your company can do.

Many companies have already reacted by putting a chat on their website.

That is a step.

But a chat is still, in essence, a storefront with a shop assistant behind the glass.

You ask and it answers.

It is available. It can be very useful. But it waits for somebody to speak to it.

It is reactive.

And we have spent a few years confusing reactive with intelligent.

From answering to acting

An agent is something else.

An agent representing a company doesn’t have to sit and wait for a question. It has an objective, and it can carry out the actions needed to reach it.

It can resolve a whole piece of business instead of explaining to the user how to do it.

Confirm an appointment.

Check availability.

Prepare a proposal.

Flag a problem before anyone has to ask.

The important difference against a chat isn’t really the technology.

It is the initiative.

One waits to be spoken to.

The other has something to achieve.

And here’s what almost nobody is watching

So far we are still picturing a scenario with a person on the other side.

That is where most conversations about agents tend to end.

And I think that is precisely where the interesting part begins.

The scenario being built is a different one:

agents talking to agents.

Your customer may never fill in your form.

It may be their agent talking to yours.

It will ask about availability, terms, prices or lead times. It will compare your offer against other suppliers. If it fits, it will be able to carry on with the transaction. If not, it will keep looking.

And all of that can happen without anyone having visited your website.

This is no longer just an idea for a conference about the future.

There are open standards trying to solve exactly this communication: letting an agent discover what a system can do, and letting agents from different organisations find each other, exchange capabilities and delegate work.

The infrastructure for it is already being built.

When this model becomes widespread, the question for a company will no longer be only whether it has a website.

It will be something more uncomfortable:

what do you answer with when the one asking isn’t a person?

A company that cannot answer will carry on existing.

But it may no longer be in the conversation where certain decisions get made.

You’ll be delegating too

There is another part we sometimes forget.

We focus a lot on the agent that represents our company, but we are also going to start delegating ourselves.

An agent that handles renewals.

That compares suppliers before you have to decide.

That cancels services you no longer use.

That negotiates or exchanges information with your suppliers’ agents.

Not a tool you have to open in order to use.

Something that works for you while you are doing something else.

In that scenario, the customer your company serves will often not be a person in front of a screen.

It will be a system acting on their behalf, following specific criteria and making decisions far faster than a person would.

You don’t win that customer over with a nice product page.

You need your company to be able to answer it.

What starting now actually means

I don’t think you need to reinvent an entire company in order to start.

What you do need is to start accumulating judgement.

And judgement is precisely what you cannot buy in a hurry later on.

Build something small and real

A bounded process, in production, not critical, but with enough contact with reality for problems to show up.

Internal pilots teach considerably less than they appear to.

Nobody behaves like a customer when they are testing something inside their own company.

The interesting learning shows up when somebody asks a question nobody planned for, uses the process in a different way, or simply expects something nobody had thought of.

Make it act, not just talk

If the first version of the agent only answers questions, you have built a chat.

That can be a good start, but it isn’t the interesting part yet.

The goal should be for it to complete a piece of business from beginning to end.

Even a small one.

Look at the agent-to-agent protocols now

You don’t need to implement every emerging standard tomorrow.

But it is worth knowing what they are trying to solve.

Above all so you don’t make architecture decisions today that you have to undo a year from now.

None of this is about getting the perfect use case right first time.

It is about starting down the road before it stops being optional.

The companies that in a couple of years have genuinely solid agents probably won’t be the ones that decided one day to bet on them.

They will be the ones that had spent time trying things, getting them wrong and fixing small problems.

Small businesses already operating this way

Nor is this necessarily a five-year horizon.

There are small businesses already working in a fairly similar way.

A customer asks a question and books directly. The appointment lands in the business’s real system and the owner finds out once it’s done.

There is no person mediating every step.

Nor does the customer need to install an app.

The interaction simply happens.

I have several examples of this kind open and accessible in Projects, in case you would rather see it working than have me explain it.

Because I think that is the crux of it.

If the storefront stops being looked at and starts receiving questions, having a storefront stops being enough.

Your company will need something capable of answering.