Is the 2:1 ratio for McKinsey more than marketing? The number of AI agents is output but not outcome and hence just a hollow number that does not matter. What are they delivering? Looking at Uber they had >70% adoption rate, burned the budget in 4 months but admitted it does not create value for customers.
I'd go further on the Tech x Human requirements of org design - the functions our businesses have now are built of necessary capabilities. The new capabilities needed are org design, work design and continuous change. That's why companies like Moderna (and others) are putting HR and IT together - but does that ACTUALLY create a new capability?
What would it take for an organisation to understand the real constraints of conceptualising the work the way they do now, in this phase of Industrial Revolution, and reimagine it for the next phase? Because that's what we're talking about .....
Does the way you reorganize burn through the human skill the whole thing depends on? "AI first, humans next" assumes the judgment is still there to add at the end. But judgment comes from practice. You get good at catching what's wrong by doing the work yourself, over and over. If AI does the first pass and people only check the end, they stop practicing, and the skill fades.
Thinking about the future in present containers is definitely one of the ways we are getting cynical. And it’s making us shut down vs being a part of building it. Thanks for adding concrete thoughts to what could be.
I’m totally with you on point 4 in particular - perhaps one of the most important questions organisations can ask themselves today isn't:
“Who do we employ permanently to do this role?”
But rather "How do we access the capability we need, at the right time, in the right way to do this thing that we need to do?”. Really appreciate your thinking and writing on this topic.
The containers aren't just structural. For most people inside these organizations, the job title, the reporting line, the defined scope of work — those are also how they understand their own value. When the acid burns through the org chart, it burns through that too.
Most organizations treating this as a design problem are going to be surprised by how personal the resistance gets. You can redraw the boxes. The harder problem is what happens to the people who spent twenty years becoming very good at what used to be in them.
Your point that the real problem is org design, not the tooling, matches what I see in project work: companies bolt agents onto the existing role map and wonder why nothing moves, because the work breakdown structure still mirrors the org chart instead of the actual work. The jobs-versus-work uncoupling is the part most leaders skip, since redrawing it touches power and comp.
One project angle worth adding: matrix orgs already failed at exactly this, work flows horizontally while accountability sits vertically. Agents just make that old fracture impossible to ignore.
Not sure what I loved most. Yours. Shelly’s. The comments. And all so much easier if you can build the organisation from scratch and not go through the pain of going from one place to another. Feels everyone will need a lot of cash in the ST.
The HI framing lands—but I'd argue it all collapses into one word: creativity.
Innovation, Imagination, Iteration—they're just creativity wearing different outfits. AI generates, optimizes, and executes brilliantly, but always from what already exists. Creativity is the one input that doesn't: the ability to want something that isn't in the training data yet.
Which is exactly why AI-first / HI-next is the right order. Let the machine do the convergent work; reserve humans for the divergent leap that sets the direction. Everyone rents the same electricity—the edge goes to whoever pairs it with people who ask the question no agent would think to ask.
The real turd on the table? Most companies are still optimizing for efficiency when the scarce resource just became creativity. I'm not really sure how younger professionals are going to learn critical thinking/creative thinking when they are so reliant on AI.
Companies will stop using AI to automate the way they currently do work but begin by utilizing AI to get work done and then add HI to augment, enhance, quality control and feeling to the AI product and solutions.
Once again, I get the message. AI and gen AI are only like the genie from Alladins Magic Lamp. The genie needs a master - and that that is a human
The tension between organizations of the past and the disruption of AI is palpable. I've observed it first hand, recently asking an executive team "If AI allows your employees to get their work done in half the time, what should they do with that remaining time?" They didn't have an answer, nor do I. The predominant work culture rewards hours and perceived effort vs. outcomes.
My second observation is that the real impact of AI at the frontlines is blurry at best. I attribute this to the fact that most employees, outside of developers, are not taking advantage of AI, whether due to lack of capabilities or incentives (see point #1). How many here would willingly use a technology that could lead to the conclusion they are no longer needed?
My hypothesis is that this will play out much like the early days of SaaS, with organizations attempting to adopt AI lagging compared to those born-in-AI. The latter are unincumbered by the tensions raised in Rishad's article and my prior points.
The point about agents ignoring old business silos is the real giveaway.A lot of org charts were never true maps of how work gets done. They were mostly maps of who reports to whom. AI has no great reason to respect that little office fiction.
Which is why the interesting question is not “where can we use AI?” It is probably, “what work actually needs doing now, who should own it, and where do we still need a human to make the call?” Otherwise we may end up with fewer people, more tools, cleaner meeting notes, and the same old confusion.Just moving at 3x speed.
Also, I am very curious whether “fractionalised employees” becomes a serious term, or just the polite name for a layoff that is still wearing formal clothes.
Interesting Post! For most of the careers I have watched, a job was a bundle, a title, a set of responsibilities, a reporting line, a compensation band, all packaged together. If that bundle comes apart, the people who do well will be the ones who can describe what they are actually good at, independent of any job description. That is a very different kind of career planning than most people have ever had to do.
Is the 2:1 ratio for McKinsey more than marketing? The number of AI agents is output but not outcome and hence just a hollow number that does not matter. What are they delivering? Looking at Uber they had >70% adoption rate, burned the budget in 4 months but admitted it does not create value for customers.
I'd go further on the Tech x Human requirements of org design - the functions our businesses have now are built of necessary capabilities. The new capabilities needed are org design, work design and continuous change. That's why companies like Moderna (and others) are putting HR and IT together - but does that ACTUALLY create a new capability?
What would it take for an organisation to understand the real constraints of conceptualising the work the way they do now, in this phase of Industrial Revolution, and reimagine it for the next phase? Because that's what we're talking about .....
Does the way you reorganize burn through the human skill the whole thing depends on? "AI first, humans next" assumes the judgment is still there to add at the end. But judgment comes from practice. You get good at catching what's wrong by doing the work yourself, over and over. If AI does the first pass and people only check the end, they stop practicing, and the skill fades.
Thinking about the future in present containers is definitely one of the ways we are getting cynical. And it’s making us shut down vs being a part of building it. Thanks for adding concrete thoughts to what could be.
I’m totally with you on point 4 in particular - perhaps one of the most important questions organisations can ask themselves today isn't:
“Who do we employ permanently to do this role?”
But rather "How do we access the capability we need, at the right time, in the right way to do this thing that we need to do?”. Really appreciate your thinking and writing on this topic.
Changing my identity and creating a feeling of empowerment/growth through the process
The containers aren't just structural. For most people inside these organizations, the job title, the reporting line, the defined scope of work — those are also how they understand their own value. When the acid burns through the org chart, it burns through that too.
Most organizations treating this as a design problem are going to be surprised by how personal the resistance gets. You can redraw the boxes. The harder problem is what happens to the people who spent twenty years becoming very good at what used to be in them.
Your point that the real problem is org design, not the tooling, matches what I see in project work: companies bolt agents onto the existing role map and wonder why nothing moves, because the work breakdown structure still mirrors the org chart instead of the actual work. The jobs-versus-work uncoupling is the part most leaders skip, since redrawing it touches power and comp.
One project angle worth adding: matrix orgs already failed at exactly this, work flows horizontally while accountability sits vertically. Agents just make that old fracture impossible to ignore.
Not sure what I loved most. Yours. Shelly’s. The comments. And all so much easier if you can build the organisation from scratch and not go through the pain of going from one place to another. Feels everyone will need a lot of cash in the ST.
Another excellent piece Rishad. Thank you.
The HI framing lands—but I'd argue it all collapses into one word: creativity.
Innovation, Imagination, Iteration—they're just creativity wearing different outfits. AI generates, optimizes, and executes brilliantly, but always from what already exists. Creativity is the one input that doesn't: the ability to want something that isn't in the training data yet.
Which is exactly why AI-first / HI-next is the right order. Let the machine do the convergent work; reserve humans for the divergent leap that sets the direction. Everyone rents the same electricity—the edge goes to whoever pairs it with people who ask the question no agent would think to ask.
The real turd on the table? Most companies are still optimizing for efficiency when the scarce resource just became creativity. I'm not really sure how younger professionals are going to learn critical thinking/creative thinking when they are so reliant on AI.
Oh wow! @Rishad Tobaccowala
Loved this
Companies will stop using AI to automate the way they currently do work but begin by utilizing AI to get work done and then add HI to augment, enhance, quality control and feeling to the AI product and solutions.
Once again, I get the message. AI and gen AI are only like the genie from Alladins Magic Lamp. The genie needs a master - and that that is a human
The hydrochloric acid visual works.
The tension between organizations of the past and the disruption of AI is palpable. I've observed it first hand, recently asking an executive team "If AI allows your employees to get their work done in half the time, what should they do with that remaining time?" They didn't have an answer, nor do I. The predominant work culture rewards hours and perceived effort vs. outcomes.
My second observation is that the real impact of AI at the frontlines is blurry at best. I attribute this to the fact that most employees, outside of developers, are not taking advantage of AI, whether due to lack of capabilities or incentives (see point #1). How many here would willingly use a technology that could lead to the conclusion they are no longer needed?
My hypothesis is that this will play out much like the early days of SaaS, with organizations attempting to adopt AI lagging compared to those born-in-AI. The latter are unincumbered by the tensions raised in Rishad's article and my prior points.
The point about agents ignoring old business silos is the real giveaway.A lot of org charts were never true maps of how work gets done. They were mostly maps of who reports to whom. AI has no great reason to respect that little office fiction.
Which is why the interesting question is not “where can we use AI?” It is probably, “what work actually needs doing now, who should own it, and where do we still need a human to make the call?” Otherwise we may end up with fewer people, more tools, cleaner meeting notes, and the same old confusion.Just moving at 3x speed.
Also, I am very curious whether “fractionalised employees” becomes a serious term, or just the polite name for a layoff that is still wearing formal clothes.
Interesting Post! For most of the careers I have watched, a job was a bundle, a title, a set of responsibilities, a reporting line, a compensation band, all packaged together. If that bundle comes apart, the people who do well will be the ones who can describe what they are actually good at, independent of any job description. That is a very different kind of career planning than most people have ever had to do.
How do you believe AI will change fiction writing and reading?