Kristie Jones treats her AI agent like a four year old. Not as an insult, as a management strategy. The same way she taught her son "no touch" before he understood why, she's learned that agents need boundaries, check-ins, and correction, just like any new hire.

Kristie, founder of Sales Acceleration Group and author of the upcoming book The Sales Leadership Gap: Leading Humans in an AI Augmented World, walked me through what it actually looks like to lead a sales team that's now part human, part automation, and part AI agent.

Why Most Sales Teams Are Only Talking About AI, Not Using It

Kristie's blunt read on the market: sales leaders are talking a good game on AI adoption, but very few are actually walking it.
She calls the hesitation "freeze mode," people waiting for the technology to settle down before they commit to anything.
Her pushback: the pace of AI isn't going to slow down, so waiting for the dust to settle just means a competitor gets the head start instead.

From Pilot to Control Tower: Managing a Hybrid Team

Kristie's analogy for what's changed: sales leaders used to be the pilot, flying the plane directly. Now the job is running the control tower.
A modern sales team isn't just a group of humans anymore. It's humans, automation, and AI agents all working together, and leading that requires a 360-degree view no single seat inside the plane can give you.
She hasn't seen a single sales leadership course that actually teaches this. Most leaders are being asked to figure it out on their own, in real time, while still being expected to hit quota.

Treat Your Agent Like a New Employee (or a Four-Year-Old)

Kristie's favorite framing, borrowed from someone else: manage your agent the way you'd manage a toddler, with simple, repeated boundaries, like teaching a child "no touch," some for safety and some just to prevent a mess.
Building an agent isn't a "set it and forget it" project. Just like a new hire, it needs a handbook, but it also needs regular check-ins, since no handbook covers every exception.
If your business is changing, and it always is, an agent that never gets revisited will quietly drift out of step with how your team actually works.

Trust But Verify

Kristie built an agent that reviews her AI notetaker every Friday at 4pm and delivers her to-do list by day. The first Friday it was supposed to run, it didn't.
Her response mirrored exactly how she'd handle a new employee's mistake: no anger, just a direct conversation about what happened and how to prevent it next time.
Her rule going forward: trust the agent to do the work, but verify the output regularly, the same discipline she'd apply to any new hire still earning full independence.

Where to Start: Update Your Playbook Before You Automate It

Kristie pushes back hard on the idea that sales playbooks are dead. They're not, but most of them are badly out of date, built once and handed to new hires without ever being revisited.
Her recommended first step for any leader building agents: go back to the actual playbook, confirm it reflects how work really gets done today, then look at it again through a new lens, asking which of these tasks could be automated.
Skipping this step means building automation around a process that's already wrong, which just makes the wrong process faster.

Two Agents Kristie Actually Built

Her first agent connects Claude Cowork to HubSpot, Granola (her AI notetaker), and Google. Every Friday at 4pm, it reviews her week's meetings and delivers a day-by-day to-do list of everything she committed to.
Her second is a research agent, built to be extensive enough that she now teaches a class on building one at Sales Gravy University.
Neither required a dedicated engineering team. She built both herself as a sales leader, which is exactly the point: if your company doesn't have a go-to-market engineer or enablement team, this becomes your job by default.

Write a "Role Brief" for Every Agent

Kristie's upcoming book outlines five leadership competencies and five AI competencies leaders will need, and one of the practical tools she recommends is what she calls a role brief.
A role brief is a formal, written document for an agent, similar to a job description but for AI: what decisions it's allowed to make, what decisions always stay human, what the output should look like, and who owns checking it.
Every agent a team builds should have one of these, the same way every human hire gets a real job description instead of a vague verbal understanding of the role.

The Two Things That Make AI Dangerous If Left Unchecked

First, AI is built to solve problems, which means if it can't find a real one, it may manufacture one to justify itself.
Second, AI essentially never says it doesn't know, and it almost never asks for a leader's opinion. It delivers a confident answer even when the underlying accuracy is shaky.
Both are reasons an agent can never be treated as fully autonomous. Someone still has to be checking the actual output, not just assuming the system is right because it sounds certain.

We've Been Here Before: What the World Wide Web Panic Teaches Us

Kristie points to the early internet as the closest historical parallel: people worried about losing control over where their information came from, the same fear now aimed at AI.
Entire households once relied on a single printed encyclopedia set as the trusted source of information. That system eventually gave way, the same way old assumptions about static CRMs and manual workflows are giving way now.
Her takeaway isn't to ignore the risk, but to recognize that fear of a major shift isn't new, and the companies that adapt early are the ones that come out ahead of it.

What AI Fluency Actually Means

Kristie is now asking about AI fluency directly in client hiring processes, and she defines it as a systems-level understanding of AI, not just knowing how to use one tool.
That includes prompt literacy (can someone actually write a prompt that gets a useful result), output evaluation (how do you know if the agent's work is accurate or has quietly drifted), and tool selection (which AI platforms actually get adopted, standardized, and connected across the team).
She's blunt with companies that call themselves "AI forward" but don't teach any of this during onboarding. Claiming the identity without building the actual skill doesn't hold up.

Your First Step to Becoming AI Fluent

Kristie's advice for any seller or leader who hasn't started: pick one tool, whether that's ChatGPT or Claude, and go build something small, like a custom GPT or a simple project.
If you don't already use an AI notetaker, that's the bare minimum starting point, and she recommends having one in place within the week.
Her closing advice is permission to fail: go make a mistake. Her own Friday agent didn't run the first time she built it, and that's a normal, expected part of learning the tool, not a sign it isn't worth doing.

"I treat my agent like a four year old." — Kristie Jones

Resources

Credits

As one of our podcast listeners, we value your opinion and always want to improve the quality of our show. Complete our two-minute survey here: thesalesevangelist.com/survey . We’d love for you to join us for our next episodes by tuning in on Apple Podcast, Google Podcast, Stitcher, or Spotify. Audio provided by Free SFX, Soundstripe, and Bensound. Other songs used in the episodes are as...