[AI SPRINT] The Hopeful Death of CRM, HRIS & EHR—Driven by AI

Last week, I gave a quick heads-up about ChatGPT’s new flexible pricing feature. It’s been causing headaches with IT teams all over, so like always, I’m trying to make it easy to deal with. Here are some immediate tips you can share:

What it is:

Now, in business accounts, when your free ChatGPT usage allotment dries up, every extra request burns through paid credits.

Quick facts:

• Credit price: $0.04 each (sold in packs)

• Team plan: seat-based limits; overages pull from a shared pool

• Enterprise: no seat caps; everyone taps the same credits

• Models: Pricey models including o3-pro, GPT-4.5, Deep Research all draw credits once limits are hit--up to $2.00 per request. 

• Visibility gap: real-time dashboards still MIA. Employees need to guess how much they use, and your bill can go through the roof if they don’t watch out

3 sanity savers:

• Teach staff to use the cheapest model that still delivers quality (see below)

• Set up billing alerts the moment you buy your first credit pack

• Designate a “usage auditor” to track weekly usage and step in if anyone abuses it

To help with this, I created the below cheat-sheet that you can use to help your team understand both which model is best to use today, and what the costs are when their allocation is used up.

Now, onto the hopeful death of CRM and redesigning your processes to be AI-first!

A CEO asked me this week, “What CRM should we buy to improve our sales activity?” It's a question I’ve answered often over the years, and on the surface, it makes sense. Sales feels chaotic, leads go dark, follow-ups get missed. Naturally, leaders think a CRM will fix that. But this question is rooted in a much deeper problem, one that goes far beyond CRM.

Most of the systems we use in business today—CRM, HRIS, ERP, EMR—weren’t designed to make human work better.

They were designed to track it.

CRMs track activity, not customer outcomes. Electronic medical records were built to log what doctors and nurses did—not to make patients healthier. HRIS tools track headcount, compensation, time off—not engagement or performance. These systems don’t move the work forward. They just document that it happened, often requiring humans to spend a good chunk of their workday entering what they just did.

The origin story of CRM says it all. The first version was released in 1987. It was called ACT: Automated Contact Tracking. That’s what it was built to do—track who you talked to, and when. It wasn’t very automated then—and CRM still isn’t.

Nearly forty years later, we’ve added dashboards, email templates, and mobile apps—but the DNA hasn’t changed. CRMs still depend on humans to feed them. If a rep doesn’t log the call, the system doesn’t know it happened. If a follow-up is missed, the system won’t notice until someone looks at a dashboard. And if your sales team is busy actually selling, the CRM becomes just another checkbox. A system to manage the work about the work.

We’ve come to accept this as normal. More growth means more tools to buy. More tools mean more dashboards. And more dashboards mean more people hired to manage the dashboards. Eventually, you have a system designed not to improve outcomes—but to keep track of who’s doing what.

But what if that’s no longer necessary?

When Intelligence Becomes Cheap, “CRM” is No Longer Needed

In my keynote talks, I often pose this question: when intelligence becomes cheap, what changes?

Historically, intelligence was expensive and systems were dumb. So we designed systems to be structured and strict. They needed clean inputs. They needed human instruction. The only way to know what was going on inside a business was to log it, tag it, sync it, and report on it. That’s why we got endless status meetings. That’s why we got admin-heavy tools. That’s why every department ended up needing its own system.

Not because we liked it—it was because we had no alternative to standardize human activity.

But now, with AI, we do. The cost of intelligence is collapsing. We no longer need humans to tell the system what happened. The system can observe it directly. It can read the email thread. It can see the drop in order volume. It can flag the delayed shipment and understand how that’s impacting your top account.

This is what I mean when I say intelligence is cheap. 

Not just that we can send a better email, or automate a few human tasks, but that we can fundamentally redesign how systems function. From something you feed… to something that feeds you.

CRM: The Perfect Place to Start

Let’s go back to that CEO’s original question. When most people ask, “What CRM should I use?” what they really want is something that gives them visibility, makes reps more consistent and productive, and ensures customers aren’t slipping through the cracks.

Totally reasonable.

But here’s what they usually get: reps spending hours logging activity, managers reviewing dashboards full of stale data, and a “system of record” that only works if everyone behaves like part-time admins.

The moment real work ramps up, usage drops off. And the most important customer information ends up scattered across inboxes, Slack messages, and someone’s memory.

You’ve probably seen this play out.

That’s the hidden cost of legacy systems: those built for humans to track their own activity. They rely on structure and compliance to function—and humans aren’t great at either when they’re busy doing actual work.

An AI-first system flips this. 

• It doesn’t ask reps to log calls—it pulls signals from inboxes, orders, support tickets, behavior.

• It doesn’t ask managers to chase down activity—it highlights accounts showing signs of risk.

• It doesn’t make salespeople create reminders—it surfaces timely nudges based on customer context.

In short, it does the job we always hoped the CRM would do, but without the overhead.

From Tracking People to Serving Them

This shift—away from tracking activity and toward enabling outcomes—is what is occurring in what we are calling “AI-native” companies. Instead of using traditional systems, they are building agents to do the mundane work. And it’s not just CRM.

Customer support AI agents can now handle common questions automatically, escalate with full context, and make customers feel heard without burning out your team. Inventory AI agents can detect shifts in SKU velocity, generate POs before you stock out, and flag exceptions before the customer calls. Logistics can become proactive, not reactive. And leadership visibility doesn’t need a dashboard—it needs a single daily signal: what changed, what’s working, what’s breaking.

This is the approach I helped build at Amazon. AI-first workflows. Where humans were informed of what is occurring, and asked for input. Not the other way around.

All of this is possible when intelligence becomes a layer in your business operating system, not an afterthought.

And that’s the real opportunity here. Not just to replace the CRM. But to reimagine how the business functions—from sales, to service, to ops, to leadership.

Final Thought

When intelligence was expensive, we built software to manage people.
Now that intelligence is cheap, we can build software that helps people manage outcomes.

You don’t need a better CRM.
You need a smarter system. One that sees what's happening, decides what matters, and helps your team act—without admin, without dashboards, without delay.

Most companies will keep stacking tools and chasing visibility.

You get to build it differently.

And if you want help figuring it out? You know who to talk to!

About Trent: Trent Gillespie is an AI Keynote Speaker, CEO of Stellis AI, former Amazon leader, and advisor on building AI-Native, AI-Enabled businesses. Book Trent to speak to your group or book a call to discuss using AI within your business.

Did someone forward this newsletter to you? If you're not already signed up, you can subscribe to AI SPRINT™ for free here.