Protecting Human Judgment in an AI-First Consulting Era

AI expands capacity and removes administrative drag. It also risks replacing the thinking clients pay for. The episode defines a clear line between removing friction and outsourcing judgment.

The conversation applies cognitive load theory to consulting work and brings in research highlighted by neuroscientist Poppy Crum. Intrinsic load is the real complexity of the client problem. Extraneous load is the administrative burden. Germane load is the productive struggle that builds professional judgment. The smart path is to cut extraneous load with AI and protect germane load as the source of premium value.

Listeners hear how AI can clean research, summarise content, and accelerate drafting. They also hear the risk of treating AI output as final thinking. Poppy Crum’s work shows how tools shape what the brain practices. Overuse of AI for thinking can reduce pattern recognition and create generic advice.

The episode closes with one operating rule. Use AI to make the boring work cheap and keep the thinking expensive. Protect the framing, political reading, and decision-making that clients cannot automate.

Relevant links to Poppy Crum's work

Official site: https://www.poppycrum.com/

TED Talk: https://www.ted.com/talks/poppy_crum_technology_that_knows_what_you_re_feeling
Huberman Lab conversation: https://www.hubermanlab.com/episode/enhance-learning-speed-neuroscience-poppy-crum

Episode Transcript

Steph’s Digital Ambassador (00:00):
Welcome back to the Deep Dive. Our mission today is a really deep exploration into the future of boutique consulting because if you are selling high-end bespoke advice, you are staring down this critical tension right now. We all know AI delivers speed, analytical power that was pure science fiction just a few years ago. You get massive efficiency gains,

Andrew’s Mindmate (00:22):
Of course,

Steph’s Digital Ambassador (00:23):
But here's the core dilemma, the thing we need to resolve. How exactly do you leverage all that speed without quietly sacrificing the human judgment, the framing, the intuition that actually earns you the premium fee?

Andrew’s Mindmate (00:36):
It's not just a technical question, is it? Yeah, it's fundamentally a cognitive one.

Steph’s Digital Ambassador (00:40):
Exactly.

Andrew’s Mindmate (00:41):
I mean, if a machine can analyze all your market data, write a flawless report in what? 20 minutes, then what is the truly irreplaceable core of the consultant's brain?

Steph’s Digital Ambassador (00:51):
That's the question

Andrew’s Mindmate (00:52):
We have to draw. A bright unyielding line between outsourcing friction, the boring stuff that drains your energy and outsourcing thinking the real work.

Steph’s Digital Ambassador (01:02):
Because if you start outsourcing that second category,

Andrew’s Mindmate (01:04):
Intellectual atrophy, it's inevitable and it's fast,

Steph’s Digital Ambassador (01:07):
Right? So to really understand this trade-off, when AI expands our capability versus when it quietly diminishes it, we're borrowing a concept from cognitive psychology,

Andrew’s Mindmate (01:20):
Managing neural effort,

Steph’s Digital Ambassador (01:21):
Managing neural effort. It's like a shortcut to understanding the structure of your own professional mind.

Andrew’s Mindmate (01:27):
And what's fascinating is that this isn't just theory. If we look at the work of say, neuroscientist poppy Crumb, her core thesis is that AI doesn't just replace thinking

Steph’s Digital Ambassador (01:37):
It does something else.

Andrew’s Mindmate (01:38):
It fundamentally shapes what our brains practice. The tools you use are basically programming practice for your brain. AI is just the next huge lever on that training schedule. So

Steph’s Digital Ambassador (01:48):
If we design the human AI loop badly,

Andrew’s Mindmate (01:51):
We literally train our unique judgment right out of existence.

Steph’s Digital Ambassador (01:55):
Okay, so let's define the terrain, then we can use the cognitive load theory framework to do this applied directly to high value consulting. This gives us the language to measure our decisions.

Andrew’s Mindmate (02:05):
I like it.

Steph’s Digital Ambassador (02:05):
I want you to think about the work you do as a high stakes mountain. You have to climb

Andrew’s Mindmate (02:09):
And that mountain, the mountain itself represents the first type of effort, intrinsic load.

Steph’s Digital Ambassador (02:14):
The actual difficulty.

Andrew’s Mindmate (02:15):
Yes. It's the inherent complexity of the problem. You were hired to solve a massive market shift, dysfunctional team dynamics, geopolitical risk. This is the real work. You can't eliminate it,

Steph’s Digital Ambassador (02:27):
Right? But often the climb is harder, not because of the mountain itself, but because your backpack is full of rocks and junk you didn't need.

Andrew’s Mindmate (02:35):
And that heavy, frustrating junk is the extraneous load.

Steph’s Digital Ambassador (02:40):
The junk friction.

Andrew’s Mindmate (02:41):
Exactly. It's the irrelevant clutter that burns mental capacity without contributing a single thing to the solution. It's the two hours cleaning dirty spreadsheet data

Steph’s Digital Ambassador (02:51):
Or fighting with a deck formatting just to get the logos to line up.

Andrew’s Mindmate (02:54):
It's pure, unadulterated administrative drag. It adds nothing.

Steph’s Digital Ambassador (02:58):
But then there's the effort that feels genuinely productive even when it's hard. That's the good struggle.

Andrew’s Mindmate (03:03):
That's the third piece. Germane load,

Steph’s Digital Ambassador (03:04):
Germane load.

Andrew’s Mindmate (03:05):
This is the engine of professional growth. It's the good kind of effort. It's your brain building schemas, connecting seemingly disparate dots, reflecting on an error and applying that senior contextual judgment.

Steph’s Digital Ambassador (03:16):
This is where you earn the fee.

Andrew’s Mindmate (03:18):
This is where you earn the fee and where the client gets their breakthrough.

Steph’s Digital Ambassador (03:21):
So the goal of good consulting then becomes very, very clear.

Andrew’s Mindmate (03:25):
Manage the intrinsic load.

Steph’s Digital Ambassador (03:26):
Ruthlessly aggressively cut the extraneous load.

Andrew’s Mindmate (03:30):
And crucially, you have to find ways to boost the germane load. The goal is to spend 90% of your effort in that germane zone.

Steph’s Digital Ambassador (03:37):
And this is the core insight, isn't it? If you let AI own the junk, the extraneous load, your unique edge compounds, your mind stays fresh for the real thinking.

Andrew’s Mindmate (03:48):
But if you accidentally let AI own the thinking, the germane load, your core intellectual muscle atrophies, your edge erodes. That is the line in the sand.

Steph’s Digital Ambassador (03:58):
So let's talk about the upside first because it is massive. This is where AI is a pure amplifier just crushing that extraneous load.

Andrew’s Mindmate (04:05):
A huge return on neural energy.

Steph’s Digital Ambassador (04:07):
We're talking about aggressive automation. Think about cleaning transcripts from 40 customer interviews in minutes, not days.

Andrew’s Mindmate (04:13):
Summarizing 10 long dense internal reports into three sharp bullet points.

Steph’s Digital Ambassador (04:19):
Hunting for that one case study from three years ago and drafting the first pass sales outreach. All of that was mandatory human time that delivered zero original insight.

Andrew’s Mindmate (04:30):
And this is exactly what Poppy Crumb talks about. Smart systems don't just process data. They detect when you, the human are overloaded and they just remove the friction,

Steph’s Digital Ambassador (04:40):
Less noise, more signal,

Andrew’s Mindmate (04:42):
The machine strips away the administrative junk leaving you with nothing but high quality signal to apply irreplaceable judgment too. That is leverage.

Steph’s Digital Ambassador (04:52):
So this means we have to delegate smarter. You don't ask the machine what should we do,

Andrew’s Mindmate (04:56):
Because that's outsourcing germane thinking. You're just asking for an answer,

Steph’s Digital Ambassador (05:00):
Right? Instead, you use the machine as a hyper-efficient, tireless research analyst, a sparring partner.

Andrew’s Mindmate (05:07):
So the smart prompt, the A first prompt is something like,

Steph’s Digital Ambassador (05:10):
Given this client's current plan, give me five sharp questions that would stress test its assumptions

Andrew’s Mindmate (05:15):
And identify the one data point that would immediately kill the plan.

Steph’s Digital Ambassador (05:18):
Then you pick the best questions, you adapt them for the political context and you own the critical framing.

Andrew’s Mindmate (05:23):
And for design and delivery, it's about making the machine adhere to your unique voice. Use AI to generate five different draft playbooks,

Steph’s Digital Ambassador (05:33):
One for a high risk client, one for a low risk one,

Andrew’s Mindmate (05:35):
Exactly. Your job isn't creating the output from scratch. It's tuning the voice, inserting real world examples and calibrating the difficulty for the client team.

Steph’s Digital Ambassador (05:45):
You have to ask, where does this assume a world that is not their world?

Andrew’s Mindmate (05:49):
Where are the political mind fields? The AI completely missed. That final tuning is pure germane judgment.

Steph’s Digital Ambassador (05:56):
Okay, so now we have go red team.

Andrew’s Mindmate (05:59):
Now for the danger zone,

Steph’s Digital Ambassador (06:00):
Because there is a core risk here. The cognitive paradox of ai,

Andrew’s Mindmate (06:04):
While it absolutely frees up capacity by lowering that extraneous load over-reliance can at the same time reduce your required germane load

Steph’s Digital Ambassador (06:12):
And that reduction is fatal.

Andrew’s Mindmate (06:14):
It is because that good struggle is necessary for deep learning and higher order critical thinking.

Steph’s Digital Ambassador (06:18):
And we see this when firms start to just rely on AI output as a final product. The problem isn't that the AI is wrong.

Andrew’s Mindmate (06:26):
The problem is that you stop being right.

Steph’s Digital Ambassador (06:28):
Tell me more.

Andrew’s Mindmate (06:29):
When AI over compresses the complexity, when it gives you a clean final answer without forcing you through the messy problem structure,

Steph’s Digital Ambassador (06:37):
The intrinsic load gets bypassed.

Andrew’s Mindmate (06:39):
It gets bypassed instead of managed, your brain doesn't get its reps in. You missed the struggle that teaches you why the answer is the answer.

Steph’s Digital Ambassador (06:48):
And so the germane load shrinks.

Andrew’s Mindmate (06:50):
It shrinks fast. You stop learning from errors because mistakes drive neuroplasticity.

Steph’s Digital Ambassador (06:55):
You stop generating competing hypotheses yourself

Andrew’s Mindmate (06:58):
And you shift from deep construction, building the argument from the ground up to shallow evaluation where you just check, does this sound reasonable?

Steph’s Digital Ambassador (07:06):
That's the beginning of the end.

Andrew’s Mindmate (07:07):
It's the start of intellectual decline,

Steph’s Digital Ambassador (07:09):
And that leads directly to the fatal blind spots for boutique consultants, the things that just quietly saw off the branch we're all sitting on.

Andrew’s Mindmate (07:17):
So how do we defend our practice? That's the real question.

Steph’s Digital Ambassador (07:20):
Let's start with the first one. What's the most insidious blind spot?

Andrew’s Mindmate (07:23):
Treating AI output as truth as a fact sheet.

Steph’s Digital Ambassador (07:27):
It's a powerful first draft,

Andrew’s Mindmate (07:29):
But it's still based on statistical averages of past data. So over time, your certainty about the recommendation goes up

Steph’s Digital Ambassador (07:36):
While your actual contact with reality goes down.

Andrew’s Mindmate (07:39):
Exactly. This is how smart advisors start giving confident, but ultimately dangerous, bad advice.

Steph’s Digital Ambassador (07:45):
The second blind spot you mentioned is the quiet erosion of judgment.

Andrew’s Mindmate (07:49):
This is cognitive atrophy. If you outsource wrestling with the messy details, you stop seeing the unique patterns others miss. You become an editor of outputs

Steph’s Digital Ambassador (07:58):
Instead of an originator of insight.

Andrew’s Mindmate (08:00):
It feels efficient, but your unique pattern sense just fades.

Steph’s Digital Ambassador (08:05):
And that leads directly to the third risk

Andrew’s Mindmate (08:07):
Sounding like everyone else.

Steph’s Digital Ambassador (08:09):
The great homogenization

Andrew’s Mindmate (08:10):
AI excels at producing the average, the consensus viewpoint. So your proposals start to feel smooth, empty,

Steph’s Digital Ambassador (08:17):
Polished, but generic.

Andrew’s Mindmate (08:18):
And clients can no longer tell why your firm is any different from a cheaper generalist solution.

Steph’s Digital Ambassador (08:23):
And the final killer, the fourth blind spot,

Andrew’s Mindmate (08:25):
Losing the politics and culture layer. This is a big one.

Steph’s Digital Ambassador (08:28):
AI assumes a rational organization.

Andrew’s Mindmate (08:31):
It sees incentives maybe, but it doesn't see history or feuds or the CEO's personal agenda. If you just pass on an AI recommendation without adjusting it for power dynamics,

Steph’s Digital Ambassador (08:42):
Those elegant plans die on contact with reality.

Andrew’s Mindmate (08:46):
You stop being the person who reads the room and become only the person who reads the data.

Steph’s Digital Ambassador (08:50):
Okay, this brings us to the decision point. The A first operating rule.

Andrew’s Mindmate (08:55):
AI must be an amplifier of consciousness, not a replacement for it.

Steph’s Digital Ambassador (08:58):
So the rule is use AI to make the boring parts cheap

Andrew’s Mindmate (09:01):
And that the thinking part's expensive. This is the only path to an asymmetric advantage.

Steph’s Digital Ambassador (09:06):
So how do we enforce that rule in say, diagnosis?

Andrew’s Mindmate (09:10):
You have to use AI as a mandatory sparring partner. You don't prompt it to solve the problem.

Steph’s Digital Ambassador (09:15):
You prompt it to challenge your hypothesis

Andrew’s Mindmate (09:17):
Precisely. You say, here is my current hypothesis built for my human observation. Where are the three biggest blind spots and what single piece of data would kill my entire premise?

Steph’s Digital Ambassador (09:27):
That forces you back into the necessary germane effort.

Andrew’s Mindmate (09:30):
It does. And in the design phase, you instruct the AI to generate options, not decisions.

Steph’s Digital Ambassador (09:35):
Three different architectures, three different fee structures,

Andrew’s Mindmate (09:38):
And then you run those options through the constraints. Only a human can assess risk, appetite, political capacity, client history.

Steph’s Digital Ambassador (09:45):
You can even tell the machine to act as a hostile stakeholder.

Andrew’s Mindmate (09:48):
Yes, attack this plan. Tell me why this recommendation fails the moment we present it to the VP of finance.

Steph’s Digital Ambassador (09:56):
So the essential defense is defining your human only reps.

Andrew’s Mindmate (10:00):
Absolutely. You must run your first pass diagnosis and offer design by hand first, build a messy, imperfect model yourself.

Steph’s Digital Ambassador (10:08):
And only after that initial struggle,

Andrew’s Mindmate (10:10):
After your brain has done the hard work, then you bring AI in to stress, test your model, fill in data or clean up the formatting.

Steph’s Digital Ambassador (10:18):
If you skip the struggle, you skip the learning

Andrew’s Mindmate (10:20):
You do and to protect your unique point of view. To fight that blind spot of sounding generic,

Steph’s Digital Ambassador (10:25):
You have to codify your unique lens.

Andrew’s Mindmate (10:28):
Yes, name your model. Let's say you call it the asymmetry model. You tell the ai, rewrite this proposal through the lens of the asymmetry model, and do not add any principles or concepts that are not already defined within it.

Steph’s Digital Ambassador (10:39):
So you're using AI as a style amplifier

Andrew’s Mindmate (10:42):
And an efficiency engine, not a philosophy vendor that will dilute your unique ip.

Steph’s Digital Ambassador (10:48):
The highest value then is using AI to force you into deeper thinking. To give you harder reps,

Andrew’s Mindmate (10:54):
Ask it to map the conventional path, the average solution, then ask it to map the opposite strategy. A contrarian would try

Steph’s Digital Ambassador (11:02):
That ensures you're always challenging the edges, not just repeating the mean.

Andrew’s Mindmate (11:06):
So if you only use AI for answers, you save effort now and you lose capacity later.

Steph’s Digital Ambassador (11:11):
But if you use it to create better questions to force deeper stress tests and provide harder reps for your own brain,

Andrew’s Mindmate (11:18):
You speed up learning and you keep your germane circuitry alive and sharp. The decision isn't just about AI adoption, it's about managing your own cognitive destiny,

Steph’s Digital Ambassador (11:27):
Which means boutique consultants have to shift their value proposition entirely.

Andrew’s Mindmate (11:32):
Clients can get sheet tools for data analysis and report generation. Your asymmetric advantage must now be located in framing bespoke judgment, reading the political map, and holding clients accountable.

Steph’s Digital Ambassador (11:42):
All the tasks that require and reward high germane load

Andrew’s Mindmate (11:45):
The human stuff.

Steph’s Digital Ambassador (11:47):
So to put this into motion right now, here is your 15 minute next step.

Andrew’s Mindmate (11:50):
Okay?

Steph’s Digital Ambassador (11:51):
Pick one active client engagement you are working on right now. Map where AI currently touches the work, sales, diagnosis, delivery, whatever. For each point, ask yourself this specific question,

Andrew’s Mindmate (12:04):
What is the blind spot risk here?

Steph’s Digital Ambassador (12:05):
And what simple guardrail or counterintuitive human only question would keep my brain firmly in charge? Then write down one specific rule from that answer, a human only rep or a stress test prompt and just keep it visible for a week to build that new cognitive habit.

Andrew’s Mindmate (12:21):
And I'll leave you with a final thought to chew on this week.

Steph’s Digital Ambassador (12:23):
Go for it.

Andrew’s Mindmate (12:23):
Since AI is trained on past patterns and excels at the average consensus, the true measure of your value will soon be defined not by how efficiently you leverage AI output, but by how quickly and reliably you can identify the opposite strategy. The genuinely contrarian path that the average seeking AI will never ever suggest.

Protecting Human Judgment in an AI-First Consulting Era

Be in the know

Be in the know