Consulting in the Age of AI
Recently OpenAI announced the launch of GPT-4o. There was a lot of release of promotional videos and one of them caught my eye. In the video, Sal Khan the founder of Khan Academy got ChatGPT to guide his son on a geometry question. (Link to the video at the bottom)
The video caught my attention because I have always been passionate about learning or more precisely I am passionate about building up human capital. Having something like ChatGPT to tutor the kids is a game changer as it can increase the learning efficiency of children, allowing them to take on more complex learning later. I have some thoughts on this too but topic for another newsletter.
Now if we extrapolate this function, besides impacting the tutoring industry which is a huge business in Asia, GPT-4o or the like is coming for the consulting profession in the longer run!
Consulting Currently
Currently, based on experience and discussions, there are two types of consulting. One is based on advisory which usually leads to a retainer, and the other is working on projects.
For advisory, clients are buying the background, experience, and thought process of the consultant. The consultant is expected to search for ideas and information, and curate and structure a solution based on the background and experience. Having a process is minimal here as compared to working on a medium-term project.
For medium-term projects where there is an output, there is usually a group of consultants coming together, working on the project through the variety of technical skills put together. There will be someone with the experience to put together the team and do the coordination effort. Such folks are usually the highest paid because they are supposed to come up with processes, design, plan, and manage them. These projects are usually charged high fees given the amount of effort required
Consulting in the Future
Projecting the development of GPT-4o and similar, here is what I foresee might happen to the two types of consultation.
For advisory, the expectation of the advisory quality will be increased. Mediocre consultants who just present solutions without putting more critical thought and curation will be eliminated from the market. Consultants that can provide a more customized and feasible solution will be highly desired. What this translates into is that for clients to judge a consultant’s ability to provide the best-customized solution, will be the number of years of relevant experience and also its portfolio of projects. The more the merrier!
For medium-term projects that have an output, given the exorbitant high fees charged, there is a good chance that companies might build up the project internally. Why? The project can be guided by using GPT-4o, from step-by-step instructions to the makeup of the teams, instead of the conductor/manager. Furthermore, team members can be put together accordingly, learn from, and apply the knowledge prompted by multimodal generative models. There will be a downside to such an arrangement but with these models being able to provide guidance, the high fees charged by building externally will be pitted against, building it and, having the skills and experience retained internally which can be the basis and foundation for further work later, rather than relying on the vendor when an audit comes. Expertise and knowledge is retained internally.
Concluding…
GPT-4o being able to prompt (back) and guide students in solving mathematical problems might just be a good glimpse into the future of consulting. We should start to ask ourselves how will it disrupt the profession. If you are a consultant, it is time to ask yourself what is your value add to your clients and work to extend that advantage further before AI erodes a lot of value out of it.
What are your thoughts on this? Does your company hire consultants? What kind of value do they provide? Will love to hear more!
Consider supporting my work! You can make a “book” donation and drop me some wisdom! Thanks in advance! :)
Here is the video that I mentioned. :)
I believe the disruption of consulting by AI will be far more significant than what has been suggested, especially for strategy consulting. Indeed, today, strategy is a commodity.
GenAI, in particular, seems to be a perfect fit to replace many aspects of traditional strategy consulting. A study done by BCG is telling: they gave GenAI to a group of consultants good and bad for a period of time. The worst performers were able to close the gap with the best performers. The best performers did not improve that much with AI. There is some levelling, some averaging happening and anyone can be as good as a BCG or Mckinsey consultant today already. Soon much better than them.
Moreover, AI is eating consultants' lunch in various ways:
1. Expertise at Scale: Traditionally, consultants bring the expertise derived from their experience working with multiple clients. Customers hire consultants for their ability to provide best practices and insights based on a wide range of previous projects. However, AI can leverage the collective knowledge and best practices from millions of cases, rather than just a few. This creates an unparalleled opportunity for benchmarking at scale, which, in turn, could become a free, easily accessible service. The expertise that clients typically pay for in a consultant will soon be readily available through AI. It is like with medecine where no radiologist can beat an AI system to detect an anomaly and interpret it on an X-Ray photo. Soon, you'll want to go only for scanning facilities that use AI, you won't trust pure human radiologist expertise. Same with consultants. You will want to use them for executing, not strategizing and advising. If this execution phase is what you call "customizing" then I agree with you: this is where consultants we have to focus. Change Management. Enablement. Driving the transformation for real.
2. Adapting to Ambiguity: One key role of consultants has always been their ability to navigate ambiguity and offer insights, even when faced with scattered or incomplete data. This was historically seen as a uniquely human skill—being able to make sense of complex situations and provide frameworks for decision-making. GenAI, however, excels at working with various types of data and can still deliver high-quality, actionable insights even when the input is messy or unclear. It doesn’t need structured data to perform well; it adapts and makes the most out of whatever information is provided, much like an experienced consultant would do.
3. Automation of Data Gathering and Analysis: Another significant component of traditional consulting is the time-consuming data gathering, analysis, and calculation processes. The senior VP sells you the strategy but really makes money sending an army of young consultants to collect and crunch data, present insights and do the storytelling. Consultants often charge premium fees for doing this work, which is necessary for providing insights and recommendations. With GenAI, all of these tasks can be automated, freeing up time and reducing the need for costly human labor. AI’s ability to quickly process vast amounts of data and perform complex calculations at scale will make this aspect of consulting obsolete.
I often hear mainly two key objections to this idea AI will replace strategy consultants:
1) GenAI hallucinates – Yes, it can generate incorrect or fabricated information. However, humans aren’t infallible either; consultants, for example, sometimes give incorrect advice or make mistakes (often without realizing it). It’s like the debate around autonomous vehicles: we know they have the potential to be much safer than human-driven cars overall. Yet, we still demand near-perfect performance from these machines, even though humans make more frequent and serious errors behind the wheel. The expectation for flawless behavior from machines, despite the greater risk posed by human drivers, is a double standard we should be aware of. When people realize the performance gap this double standard will disappear.
2) GenAI is not accurate or predictable – A common example is the “Strawberry R’s” viral trick, where LLMs fails to count the number of ‘R’s in the word “Strawberry.” While this flaw has been addressed, I think, it raises the question: if we can’t trust an AI for something simple, can we trust it at all? But again, humans are not always accurate or precise either. For instance, if you ask an American to draw a $1 bill, most will struggle with the details by a lot —even though they use these bills everyday, for years. Does it mean you can't trust them? So why expect a system designed to mimic human intelligence to be perfect in every detail, when humans themselves aren’t? Double standard again, that is masking the true performance gap.
In short, while GenAI may not be a perfect fit for highly transactional, production-based scenarios in some industries, where you want extremely accurate and predictable data outputs, which is the world of traditional automation machines, it has the potential to completely transform the white-collar world—especially consulting. GenAI’s ability to provide expert-level insights, adapt to ambiguous inputs, and automate data-driven tasks makes it an ideal replacement for many traditional strategy consulting roles, like for doctors and legal teams.
Rather than augmenting strategy consultants, AI will likely render much of their work obsolete, offering a more cost-effective and efficient alternative to businesses seeking advice and solutions. The way out for these consultants beyond the few "consultant in the loop / human oversight" that will remain, is change management, implementation support, taking the customer by the hand. Not strategizing.
"Mediocre consultants who just present solutions without putting more critical thought and curation will be eliminated from the market. Consultants that can provide a more customized and feasible solution will be highly desired."
Absolutely true.