For the last few years, we’ve been fascinated by what artificial intelligence can tell us.
The next few years may be about what it can do for us.
That is the idea behind AI agents. Instead of simply answering a question, an agent can be given a task, a set of parameters and permission to get on with it.
Think less “Which flight should I book?” and more “Find the best flight within my budget and preferences, and handle the details.”
It sounds like a small difference. It may turn out to be a very important one.
A workforce in your pocket
Imagine a plumber whose agent answers enquiries and schedules appointments while he is underneath someone’s kitchen sink. Or an entrepreneur whose agents handle customer questions, monitor inventory and prepare marketing campaigns.
Or an investor whose agent reads company reports, follows markets and organizes information in the background.
For most of history, doing more meant hiring more people. AI agents could begin to change that equation, giving individuals and small businesses access to capabilities that once required a team.
Follow the money
There is an interesting investment story here too.
So far, much of the attention has gone to the companies building AI and supplying the computing power behind it. But if agents become commonplace, an entire economy will develop around them.
They will need software and data, identities and permissions, payment systems and, importantly, security.
Giving AI permission to summarize your emails is one thing. Giving it permission to spend your money is another.
So the opportunity may extend well beyond the companies building the smartest models. Cloud infrastructure, payments, cybersecurity, identity and specialized software could all become part of the ecosystem that allows agents to operate.
We’ve seen this before. The internet created far more than internet companies. The smartphone created far more than phone manufacturers.
AI agents may create an ecosystem of their own.
The other opportunity
But there is another side to the story.
The opportunity may not simply be investing in the companies building AI. It may be learning how to use it ourselves.
At Miravest, we decided to find out.
In August, the Miravest Global Balanced Fund created two AI agents to support the portfolio manager.
The first works daily. Each morning it produces a concise seven-point CIO brief: what changed overnight, what matters for the portfolio today, risks and opportunities, upcoming events and where attention may be needed.
The second takes a longer view. Once a week, it reviews the portfolio and investment environment — performance and positioning, changes in the market backdrop, asset classes and sectors, risk exposures and where our convictions may need another look.
Both operate within clearly defined parameters and constraints. Neither makes investment decisions. That remains the responsibility of the portfolio manager.
But our first impressions have been impressive.
What has surprised us is not any single thing the agents do. It is how much useful work can already be done in the background, consistently and within clear boundaries, leaving more time for thinking and making decisions.
And perhaps the most interesting part is the sense that we are only scratching the surface.
As agents improve, gain access to more tools and become better at working with other systems, what feels novel today may simply become part of how we work.
For investors, that makes the agent economy worth watching. But our experience also suggests something else: the interesting question may soon move beyond what AI can tell us, to what we are comfortable asking it to do for us.
Who’s your agent?
This communication is provided for information purposes only and does not constitute investment advice, an offer, or a personal recommendation. Past performance is not a guide to future results. Prospective investors should review the fund’s full documentation and seek independent advice appropriate to their circumstances before investing.