Just Because I Can - A Detour About Resourcing

People Are Not Resources, Can AI Agents Be?

On the previous post of this series, I said that I would talk about my current tech stack and AI tools. I felt like taking a detour on #justbecauseican to talk about the topic of resources.

After reading about the onslaught of AI replacing humans [1][2][3][4][5][6], and becoming aware that these trends are impacting the jobs of some of my outstanding former colleagues, I felt that I had to opine on the topic.

Those who worked with me know that I led a campaign -- perhaps a career-limiting move -- against the word “resources” as applied to human beings. My lemma “People Are NOT Resources” was well known within the corporation.

In my most recent corporate roles, I helped on preparing the company for exit by combining acquisitions with the development of a leading-edge platform and new products, while gradually migrating customers to that platform. At one stage, I managed the integration of teams from six acquired companies across eleven global development centers. A key mandate was to consolidate and scale operations by reducing the number of centers. Such directives were standard before COVID reshaped practices, making remote and distributed work widely acceptable.

Because such acquisitions usually involved distressed companies, we had to deal with an exodus of their leadership and top talent. We were left with a core team of what I would call "true believers.” After all their hardships, this core team was committed, knew their business like no one else, and were willing to stick with us because they believed in their mission and, more importantly, they had a strong feeling of responsibility and ownership towards their customers.

This was a frequent directive: “Move these ten resources to a lower-cost country. I'll even give you twelve replacement resources there”.

But PEOPLE ARE NOT RESOURCES.

Not like chips in a casino. People are not fungible objects. People are not expendable assets -- at least not in knowledge-centric professions. Even that server at a restaurant who knows your name and preferences is not a "resource" and might very well be the reason why people patronize the place.

The intrinsic knowledge that someone has about their domain; their customers; their software (which is rarely well documented); their operations, is not easily transferable to an outsourced team in a lower-cost country. Besides, the good and talented people in a lower-cost location also want to work on the latest and leading technologies, not on someone else's legacy. And trust me, it doesn't take software too long to enter the "entropy zone", to be legacy.

So now, the trend is to outsource the work to AI Agents. Will it work? Can AI Agents be these magical resources?

Are Agents the Solution?

I am a strong believer in Agents and Agentic AI. Currently, I am an advisor at BDB.ai where they have developed a powerful Agentic AI solution. We have found that agents bring an incredible capability for semantic understanding of concepts in data. Some of the auto generated queries and data pipelines are astonishing; it would have taken a human being days to develop some of these queries.

Since I have been retired for some time, I didn't have the opportunity to work on agentic transformation initiatives. I suspect that some decisions about replacing people with agents will be ill-conceived if based on optimistic assumptions about the fungibility of resources.

I am going to extrapolate from previous experiences of "replacing" people with people and contrast them with what could be done with AI. My hope is that we can open a discussion space for what works when replacing people with AI.

1. Passive Ressistance - Why would I train my replacement?

AI is very good at things that it has seen before. But each company's domain knowledge, data, processes, workflows, customers, regulations, etc. are unique, and usually, are not cast in documentation.

People-2-People: When a person was asked to train his or her replacement at another location, one could see a certain degree of passive resistance. There were some incentives but to make the financial model work, these were not usually that generous. In some cases, it was not easy to find people anywhere with experience or interest on legacy technologies. Can you say AS400? In most cases, it took more time than anticipated and when the person left, we still lost a fair amount of knowledge.

People-2-Agent: Will it be different? For sure, similar challenges, some advantages:

• AI can absorb knowledge at a much faster pace. It can read the code base, the bugs history, the backlog, the customer history, etc. Still, much of the knowledge was on people's heads and not scribed for AI to use.
• After absorbing the available knowledge, an AI agent can develop a comprehensive plan to interview the experts and fill the knowledge gaps.
• An agent will have fewer reservations on working on legacy technology than a person. Yet, there might not be any publicly available training material for less common technologies, for example proprietary scripting languages (I am guilty of a couple of those).
• Furthermore, AI can facilitate the conversion from an older technology to a modern one.

2. Better to amplify than replace?

Can a company get more value from re-training people and amplifying their capabilities with AI?

People-2-People: We found that people working on older applications felt stuck supporting what was an important customer base, and didn't have time to learn and work on trending technology. As our transformation strategy included developing a newer, leading-edge platform, we engaged the legacy developers such that half their time was on support and half the time was on the new solution. The result was invigorating to the "legacy" developers, and we gained significant value from transferring their highly specialized knowledge to the new solution. We were even able to combat passive resistance and increase overall knowledge across various development centers.

People-2-Agent: I never was able to work on a product where the capacity of my teams was greater than the demand. I would have been afraid to work on such a product as it probably meant it was dying.

Rather than replacing a person with an agent, like for like, could it bring more value to amplify the capabilities of that person with agents, to make them 10X more productive and be able to address the product backlog at a faster pace? Of course, the person needs to be willing to learn and embrace the transformation.

Conclusion

I am not against leveraging technology to meet company objectives, and sometimes that means displacing people. We need to be business-minded too. In most cases, incentives and objectives for product leadership are largely tied to the same compensation plan as the CEO or CFO.

But people are not resources in the fungibility sense. We should discuss what the sensible and responsible ways to leverage AI are -- ways that add greater overall value, not just reduce costs.

What are your success stories? What is working for you? What strategies are you following? Feel free to comment, disagree or propose. We can all benefit from our collective knowledge and experiences.

Citations

• [1] How close is AI to replacing product managers? Closer than you think. Reddit
• [2]. (2025, August 6). 48 jobs AI will replace by 2025: Check if yours is at risk. WINSSolutions. Winssolutions
• [3] (2025, August 26). Will AI replace humans in user research? Outset.ai. Outset
• [4] (2025, January 27). AI in the workplace: A report for 2025. McKinsey
• [5] (2024, June 20). AI is replacing human tasks faster than you think. CNN Business
• [6] (2025). The AI transformation of product innovation. ScienceDirect

This post is part of the ongoing series on Just Because I Can.