The Manager as an Agent: A Future Closer Than You Think

Growth Without Hiring

Peter Guagenti, CEO of agentic AI platform Everworker, tells a story about wealth management. A company tripled its assets under management without increasing its headcount, by using AI to tackle routine tasks. The biggest one is producing investment reports and valuations, which Stanford graduates used to do.

Guagenti says that AI is absolutely going to destroy jobs, chief among them the Business Process Outsourcing roles that are a staple of MSBC, the company I work with. He also argues that the terms AI-first and Agentic AI are marketing spin. It is simply good practice when assessing new tasks to ask if there is a way to automate them.

Guagenti spoke on a panel at last week’s AI Summit, part of London Tech week. The moderator, an analyst at Omdia, broke protocol to disagree with him. She believes that human-in-the-loop is how AI will evolve, based on her conversations with enterprises. The two clashed a couple of times, but who is right?

The Rise of AI Workflows

Agentic AI was the standout theme at London Tech Week. Most of the pitch booths hosted companies offering some type of agentic workflow. Everworker went further, claiming to act as a central “brain” that builds agents for you.

To become agentic, agents need orchestration. This means a digital manager that assigns the right task to the right agent at the right time. Platforms like IBM watsonX are a no-code orchestration layer, making automation possible without developers. What used to be complex IT, is supposed to become something anyone can do.

An orchestration layer assigns and assesses tasks and compiles the results into a complete process. When asked when we’d have the first autonomous company, the panel moderator replied decades. Guagenti said 18 months.

He gave the example of real estate management. This is a low margin business, still struggling with the rise of work-from-home. Venture capital companies are going through a lean patch in terms of generating cash. Their latest wheeze is to buy up and automate real estate managers and extract the savings as dividends.

Real estate management is a series of repeating steps. You collect rents, advertise availability and sign standard agreements. If you need a human to perform maintenance or collect debts, you book one but don’t need to employ them. It is easy to see how AI could run these operations.

Yet companies must grow to survive. In real estate that means spotting up-and-coming locations, determining the trustworthiness of counterparties, and leasing or buying new properties. Market research can be automated, while negotiating may require a human until there is an agent on the other side. Full automation will require a range of services in different industries to be automated. We should expect human, autonomous and semi-autonomous workflows to coexist until that point. 

Is Agentic AI, AGI?

A further consideration is who sets the company up. Venture capital companies are run by humans for humans. It is their managers who decide to automate an industry. It is entrepreneurs who will set up companies as autonomous operations.

The point at which AI can decide to start a company sounds a lot like AGI. There was a panel at the Summit about this. Two of the members had no clue about AGI and the others shared my view that we cannot code creativity when we don’t know what it is. The LSEG’s Director of Engineering said it was enough of a challenge to get the data right, let alone coding intelligence.

Marc Andreesen says a16z’s core thesis is “Incumbents will be nuked; everything will be rebuilt.” Azeem Azhar in his Exponential View newsletter, says that senior executives he has met at dozens of public companies are “sleepwalking into a series of Blockbuster-Blackberry moments”.

Azhar believes industry changes will unfold over the next two decades. That gives plenty of time to adapt but you should be starting today. Once your suppliers and customers have agents working, they will expect you to have agents to interact with them. This will be as important as having a human-in-the-loop for other relationships.

I’ll leave you with one more quote from the AI Summit. Danny Quinn, Managing Director of DataVita said “If we gifted people GPUs they wouldn’t know what to do with them.” That is a fair reflection of where many businesses are today. That is why the next evolution of MSBC is offering Workforce AI Training for companies worried that workers are reluctant to change. If you’re passed that point, we offer Workshops for quick wins using off-the-shelf tools.

Please hit reply and ask for more details.

Questions to Ask and Answer

1. How are my customers and suppliers experimenting with AI?

2. When do they expect to be able to interact with my AI?

3. What new ventures could I start from scratch?

Here are 3 ways I can help:

1. Hit reply to ask about AI Workshops.

2. Explore AI use cases.

3. Book a consultation with an AI expert.