How to build frontline trust in AI agents (and why coaching beats prospecting as the starting point)

Overview: Resi’s AI adoption started with coaching rather than prospecting because the first priority was building frontline trust. Positioning AI as a “coach in your pocket” helped frame it as a source of value rather than surveillance, while continuous rep feedback strengthened trust over time. From there, a human-in-the-loop model kept empathy and intuition with people while shifting tedious work such as account research and first-pass personalization to agents. Centralized workflows and clear safeguards then made it easier to scale adoption without adding friction.

In a recent conversation with GeekWire, I shared how we approached AI adoption at Resi and why we deliberately started with coaching instead of prospecting. For us, the question wasn’t simply where AI could automate the most work. It was where we could create enough value to earn trust from the frontline reps who would use it every day.  

Most teams assume the first place to deploy AI agents is prospecting: write more emails, build more lists, push more activity. I’ve found a different starting point can unlock everything that comes after it.

At Resi, our entry point for AI agents was coaching.

That choice wasn’t accidental, and it wasn’t about chasing a quick productivity spike. It was about earning trust from the people who would have to live with these tools every day — frontline reps — and giving them something they could actually believe in long-term.

Why Resi started with coaching, not prospecting

When I think about sellers, across roles, segments, and experience levels, one pattern shows up again and again: reps are usually their own harshest critic and their worst cheerleader.

They’ll replay one bad call all day and miss the things they’re genuinely great at.

That’s where AI coaching becomes powerful. Its biggest value isn’t that it’s new or flashy — it’s that it’s objective. It can give a rep a true read on their calls and “do both the job of the critic and the cheerleader at the same time.”

Just as importantly, coaching was something our reps could grab onto and believe in. We weren’t trying to win a race to automate the most; we were trying to build durable adoption. Coaching let us lead with a clear rep benefit, then expand the agent footprint from a foundation of credibility.

“Coach in your pocket”: reframing AI from surveillance to value

If you’ve ever rolled out call recording or conversation intelligence, you know the immediate reaction many reps have:

This is surveillance. This is big brother.

So we got very intentional about positioning. When we introduced call recording (back in 2022), we anchored it to a rep-first value story: we want to give you a coach in your pocket.

Why? Because enablement teams can’t be on every call. A manager can’t sit beside every SDR. But an agentic model can.

That framing matters. It turns AI from something happening to reps into something working for them. And without that shift, even good technology will struggle to earn the right to stay.

Trust and value compound over time

Frontline trust doesn’t appear because you bought a tool. It’s built through repeated cycles of value delivery.

In my experience, adoption comes down to two things: trust and value, and “those both compound over time.”

We’re now years into this journey, and the reason trust exists is simple: we didn’t ship something once and walk away. We refined it continuously based on rep feedback. Over time, that consistency signals something important to the field:

This isn’t a flash in the pan. This is part of how we work.

And when reps believe you’ll keep improving the system with them (not at them), they’ll keep leaning in.

Human-in-the-loop: what stays human vs. what goes to AI agents

I’m a big believer in human-in-the-loop. We hire people for a reason. Reps bring two things that are core to how we sell:

• Empathy ‍
• Intuition

Those aren’t side quests in a sales motion; they’re the point. So, I don’t want to designate those activities to an agent.

What I do want is to shift the tedious parts of the process — the work that doesn’t require empathy or intuition — over to agents so humans can spend more time doing what they were hired to do.

In practice, that’s looked like deploying agents to support prospecting prep, including:

• Creating account briefs
• Helping reps understand what matters going into outreach
• Doing a first pass at personalization in some cases

The goal isn’t to remove the rep from the narrative. It’s to give them a stronger starting point, so their human touch is more effective.

This matters even more in niche markets where standard enrichment breaks down. In segments like house of worship, you can’t assume the information you need is neatly packaged in the usual places, “Pastor Steve’s probably not on LinkedIn.”

Before agents, an SDR might spend an hour or two digging through a website to find basic context. Now, we can prepare a brief that reduces that time drastically. The payoff isn’t just efficiency, it’s role integrity. Reps can spend more of the day calling, booking meetings, and building relationships, instead of being “pigeonholed into a data analyst role for two hours a day.”

Transparency and organic feedback as an ongoing discipline

The human/agent split isn’t a one-time design decision. It’s a system you’ll keep tuning.

One principle I repeat constantly is: transparency builds trust. If reps don’t understand why you’re rolling something out, the what feels ambiguous — and ambiguity is where skepticism grows.

So as new capabilities come online, I focus on sourcing organic feedback:

• How is this working for you?
• What do you like?
• What don’t you like?

And I try to keep that feedback grounded in outcomes, not vibes. I joke with my team: I don’t really care how you feel about my agents, I want to know what my agents did for you. If we need to adjust them, that’s fine. The point is to keep tuning the system around what helps reps execute.

That loop — transparent intent, real usage, honest feedback, visible iteration — is what keeps adoption healthy without turning every rollout into a culture battle.

Eliminating “swivel chair” workflows

Another trust killer is friction.

If a rep has to bounce from tool to tool all day: research in one place, sequences in another, coaching somewhere else, you’ll feel it in adoption. It’s cognitive load, it breaks focus, and it makes the agent experience feel bolted-on rather than integrated.

We centralized our execution model in Outreach back in 2022. That’s allowed us to run research agents there, then have agents take action on the resulting insights (like placing what they found into the sequence for the rep) while keeping the rep in a single environment for day-to-day execution and coaching.

I strongly dislike “swivel chairs.” If I can eliminate the need for a rep to go system to system in a normal day, that’s my goal. The more cohesive the workflow, the more likely reps are to use it, and the more likely the agent becomes a real teammate instead of another tab.

Advice for GTM leaders: set safeguards, then let people fail forward

If you’re scaling AI agents in a GTM org, here’s the biggest mindset shift I’d recommend:

Failing is okay as long as you’re moving forward.

The way I think about it is the bowling bumpers analogy. When you put the guardrails up, a kid can throw the ball as hard as he wants inside the safeguards. He can also roll it gently. Either way, the system is designed to keep experimentation from turning into catastrophe.

That’s the play: set safeguards, then get after it. Within guardrails, you’re likely to “fail forward” because every outcome teaches you something:

• This worked — do more of it.
• This didn’t — why not, and what changes next time?

And don’t design those guardrails in a vacuum. Talk to the reps doing the day-to-day work. I never want to be in a place where I’m the smartest person in the room for every decision. I’m not on every call. I’m not making every cold call. The people in the motion can tell you what would make the most meaningful change in their day, then you can prioritize what will have the most impact.

Frequently asked questions about frontline trust in AI agents 

Does AI adoption have to start with prospecting?

No. At Resi, we started with AI-powered coaching on purpose. Coaching created a low-friction, high-value entry point that helped reps build trust in the tool first—before asking them to rely on agents for higher-stakes parts of the revenue motion.

How do you get sales reps to trust a new AI tool?

Start by positioning it as value, not surveillance. For us, reframing AI coaching as a “coach in your pocket” helped avoid the immediate “big brother” reaction that often comes with call recording. Then reinforce trust by being transparent about the why, and refining the system over time with organic rep feedback.

What tasks should stay human vs. go to an AI agent in sales?

Here’s a breakdown:

• Keep with humans: work that requires empathy and intuition — calling, building relationships, booking meetings, and the human touch in messaging.
• Shift to agents: tedious prep work that doesn’t require empathy — account research, account briefs, and first-pass personalization support.

What is a “swivel chair” workflow, and why does it matter for AI adoption?

A “swivel chair” workflow is when reps have to rotate between multiple systems all day to do one motion — research in one tool, execution in another, coaching somewhere else. It creates friction and reduces adoption. Centralizing workflows into a single pane of glass helps agents feel integrated into rep reality instead of added overhead.

How should GTM leaders approach experimentation with AI agents?

Set clear safeguards, then let teams experiment within them. Expect some failure — and treat it as learning. Combine that with continuous, sourced feedback from the people doing the work so you’re adjusting agents based on real day-to-day impact, not top-down assumptions.