AI WON'T REPLACE JUDGEMENT. IT WILL MAKE IT MORE VALUABLE.

Last month, a Wednesday was the perfect NYC September day. Sunny with a light breeze. I was in the city to attend the Primary Venture Partners NYC Tech Summit and the content on stage was so good that I barely made it out of my seat.

That is not always the expectation at a full-day conference. There is usually a little bit of excellent content and a little bit of “we’ve all seen this slide before,” where I excuse myself and go find a corner where I can open my laptop.

Not the case at Primary.

The day was, unsurprisingly, full of AI: models, compute, agents, infrastructure, applications, and the race to build the next category-defining company.

AI can generate options. It can summarize information, draft the email, flag anomalies, write code, and accelerate work that used to take hours or days.

But it cannot decide what is worth doing in the first place. And everyone on stage agreed on this. AI cannot reliably tell you which tradeoff is right for your business, when to take a risk, what to stop funding, or whether a shiny new capability is actually connected to a durable business model.

That is still our job and not only is it sticking around, but it’s also what the most successful people in the future will do best. Use their judgement.

Here are four ideas from the Primary Conference that have stayed with me.

1. Human judgment is not going away

Jeanne DeWitt Grosser, COO of Vercel, made a point that felt particularly grounding amid all the AI conversation: human judgment is not going away.

When everyone has access to more information, more tools, and more ways to produce work quickly, the differentiator is no longer simply output. It is discernment.

What should we prioritize? What is “good enough” to move forward? Where do we need rigor? What should be automated, and what should remain intentionally human? Which signal matters, and which is simply noise dressed up as urgency?

These are not small questions. They are operating questions.

The same is true in procurement. A tool can identify potential savings. It can compare supplier proposals, summarize contract terms, and surface usage data. But it cannot fully understand the strategic context behind a vendor relationship, the internal politics that could stall a change, or the difference between a low-cost decision and a high-value one.

2. We are entering an allocation economy

Dan Shipper described an “allocation economy,” and I have been thinking about that phrase ever since.

For a long time, the advantage in business was access: access to capital, talent, information, distribution, or expertise. AI is changing some of those equations quickly. More people now have access to capabilities that once required large teams, deep technical specialization, or significant budgets.

So, the question becomes: How well do you allocate?

How do you allocate capital when there are endless tools, vendors, experiments, and initiatives competing for it?

How do you allocate your team’s attention when everything can feel urgent?

How do you allocate decision rights so the right people can move quickly without creating chaos?

And perhaps most important: how do you allocate away?

The best managers have always been good allocators. They know what their team should own, what they should delegate, what they should automate, what they should buy, and what they should stop doing altogether.

That skill will only become more important.

At Procure and Prosper, we see this constantly. Many companies have an allocation problem. Money is being spent, but without a clear point of view on which vendors truly support the strategy, which tools are duplicative, which “small” purchases are creating future complexity, and where a dollar could create more value elsewhere.

Better allocation is not about saying no to everything. It is about making the yeses count.

3. Creating problems for yourself can pay off

Matthew Prince, co-founder and CEO of Cloudflare, shared that part of Cloudflare’s success came from creating problems for itself and then solving them.

I loved this.

The strongest companies do not wait until they are forced to confront a constraint. They anticipate the next hard thing: the scale they want to reach, the customer expectation they have not yet met, the complexity their current operating model cannot support, and they begin building toward it before it becomes a fire drill.

That is a very different posture from reacting to what is already broken.

For operators, this is the work. It is asking:

• If we doubled tomorrow, what would fail first?

• What process are we tolerating because it is “fine for now”?

• Which vendor, system, or workflow will become expensive once we scale?

• Where are we relying on individual heroics instead of building a repeatable capability?

Procurement is often brought in after the pain is visible: when spend is out of control, contracts are scattered, stakeholders are frustrated, or a vendor decision needs to happen yesterday.

But the real value is earlier. It is in helping a company create the right productive problems before growth creates the painful ones.

4. Transformative technology is not the same thing as a great business

This may have been my favorite reminder from the day. Mostly because it agrees with my thoughts on how OpenAI and Anthropic are positioned in the market.

The temptation right now is to confuse excitement with strategy. To buy the AI tool because it is the AI tool. To launch the feature because competitors are launching one. To call something innovation because it has a new interface and a lot of momentum behind it.

The better question is: What business outcome does this meaningfully change?

If the answer is not clear, the technology may be impressive, but the investment may not be.

Not to mention that the technology could be game-changing, but the companies themselves may not have a line of sight to profitability. (I said what I said.)

The through-line: judgment compounds

The big takeaway from the day was not that AI will make work less important. It was that the quality of our decisions will matter even more.

The leaders who win will be the ones who can see past the volume of options and make smart, timely, disciplined choices about what deserves resources.

The managers who win will be the ones who can allocate talent, attention, and capital with intention.

And the companies that win will not simply adopt transformative technology. They will build the judgment, operating discipline, and commercial clarity to turn it into something durable.

That feels like a good reminder heading into a week full of priorities, possibilities, and probably more tabs open than any of us would like to admit.

Procure well. Prosper more.