2025 is the year of Agentic AI

Well, it's already February and what's become clear is that Agents are a thing. 

Agentic AI is going nowhere. It's here to stay. And this landscape is moving super, super fast.

So much has happened already in 2025. Only a few short weeks ago ‘Operator’ was launched by OpenAI. What may have appeared relatively inconsequential at first glance, this was in fact our first glimpse at an Agentic future.

Not just another tool. Not just another shiny new app. No, ‘Operator’ was a indicator of what's coming next.

We've moved into a new phase of the evolution of AI. Chatbots are no longer simply answering our questions or automating some regular tasks.

• Now it's all about autonomy.

• It's about having agency.

• These systems are designed to act independently.

Agents solve problems and can execute tasks that are complex without constant input from us humans.

We can give Agents a problem to solve, in an ambiguous environment and they independently work towards achieving their goal.

With OpenAI's ‘Operator’ the Agent has its own browser and it can use the web; any website on the internet, without any need for special integration, website updates or additional coding.

The Agent can consume any existing site and it can perform tasks towards a given goal. So in the example that we saw it booked some travel and it did some shopping. The Agent just went off and did those things on its own, making decisions as it went.

We are no longer just correcting grammar. This is no longer simply you running some predefined workflow.

The Agents are operating on their own to achieve something.

Consider another scenario that might be interesting.

It's someone's birthday,

The Agent would know that it's their birthday, would buy them a relevant gift; as it has some information about them, and it sends that gift to arrive on time.

It's making all those decisions on its own.

Only a week or so ago, we saw the latest agent land from OpenAI.

‘Deep Research.’

Now, this agent can be given a prompt and it will take the time it needs to discover and analyse hundreds of sources of information that are out there online. Following each source it reasons what to do next. When it believes its analysis is complete the Agent uses the information that it has gathered to write a comprehensive report.

The demo that we saw showed the Agent take tens of minutes to reason its way across the web deciding which websites to use and which sources were important. It then generate a report with its findings.

This would have taken a knowledge worker many, many hours to complete.

You can see where this is going…..

‘Deep Research’ has shown us the potential of what Agentic computing can do. It is a whole new way of working, a whole new way of thinking about how computers can work and what software really means.

And let's be honest, this is going to be massively disruptive.

This is going to reshape entire industries.

2025 is going to be the year where early adopters in Agentic AI will gain a  significant advantage.

Now is the time that we should be exploring. 

Now is the time that we should be experimenting.

How can we integrate this incredible new technology into our businesses and into our lives?