The importance of clear communication during an AI rollout

Bringing AI into your support organization will transform the day-to-day jobs of everyone in it, both support reps and managers. For those who have spent years honing their expertise, there’s potential for it to feel destabilizing. Left alone with that feeling, they can disengage or resist the tools meant to help them.

The solution is communicating what’s happening with your teams early and often, and making people feel like active participants in the change rather than spectators watching it happen to them.

They need to know three things:

• What happens to the extra capacity AI frees up
• How the support job itself is changing
• What’s expected of them going forward

Here’s how to have those conversations.

AI taking on more work can understandably leave people wondering what will be left for them. You need to be direct with your teams about how much extra capacity you believe AI will free up, and what you plan to use it for.

What you use the capacity for will differ from business to business. For us, one of the things we did was open up new channels. We started doing video calls, which is something we'd never really done before. And if extra capacity leads to being overstaffed, be honest about that too – like whether the plan is to hold headcount flat and not backfill through attrition, or reduce what you spend on outsourced support. Whatever it is, communicate the plan.

You might not have all the details of how the teams will spend their time yet, but having an open and honest conversation with them about what you do and don’t know is still helpful. Otherwise, people will start to fill in the blanks themselves. When we’ve been direct with the teams about what we didn’t know, the feedback has always been some version of “thanks for being straight with us.”

Nobody likes being told that their job is changing beyond their control. But we’ve found that people who help shape that change show up differently than people who just have it handed to them. As you’re working through an AI rollout, make an effort to bring your teams into the process so they feel just as invested as you.

AI will take on the majority of support queue work, meaning your human reps will move to focus on the hardest outlier queries and managing the AI system itself. Start by designing your AI Agent deployment so your teams can help with it, and show them what they’ll be doing day to day: QAing, shaping what the customer journey looks like, and figuring out the handoff from Agent to human.

In the traditional model, support managers focused on things like support volume, queue wait times, and productivity metrics like CSAT, response time, and number of cases handled.

But as support reps’ work moves to managing and optimizing the AI Agent instead of only working the queue, the nature of their managers’ work changes as well.

To understand how AI works, managers need to be close to the work itself. So managers become “player-coaches,” a hybrid role that’s part strategist and part operator. They need to be able to roll up their sleeves to analyze the AI Agent's performance, refine content, and debug handoffs, and also coach their team through a new way of working.

People will be more engaged the more they understand they’re part of the Agent’s ongoing success, rather than bystanders to it. Rally them to look for ways to improve its performance, like digging into the queries it couldn’t resolve and finding fixes. At Fin, we designated “out-of-inbox” time to encourage and codify this practice.

It’s also important you foster a culture of experimentation and discussion. AI is evolving fast, which means best practices for how to use the technology are constantly evolving too. Your teams will benefit if they feel empowered to develop and test their own ideas, share what they're learning, and pick up on what others discover. Create lunch-and-learns or after-work hack nights where reps can demonstrate what they built, or how a tool helped with their own work. And give your evangelists a platform to document and share their findings. Peer-to-peer proof is more convincing than leadership saying the same thing.

As the work changes, your expectations of your teams will change alongside it. Be explicit about what those expectations are so nobody has to guess.

Your reps need to learn a whole new way of working in managing an AI system, and the remaining volume-based work that now reaches them will be difficult and ambiguous by nature.

They will need to upskill, so be direct about where that needs to happen. For example:

• Technical fluency: Your reps need a working grasp of the more technical parts of what they support, the kind once left to a support engineer or handed to product. They don't need to be experts, but the baseline has to move up.
• Navigating ambiguity: With no documentation to go on, the job becomes pinpointing the problem and then chasing down the answer. This could be from the product manager, or the engineer who shipped a recent change.

Managers need to stay close to the work to understand where the biggest areas of difficulty lie, because when someone says the work is now “too hard,” there can be some truth in that. The alternative is that people are resisting adoption of new tools or ways of working, and managers need to be able to tell the difference.

Managers should acknowledge difficulty honestly and lean on the tools and training that make the harder work manageable. The goal is to help their team build the confidence and capability to work differently, at a pace that can be maintained.

And in cases of non-adoption, managers should:

• Challenge preconceived notions: In my experience, tenured teammates can be some of the most resistant to change as they had established workflows. Highlight ways AI tools could be integrated into these and encourage a willingness to try again, especially as tools meaningfully improved.
• Make AI literacy and usage an expectation of the role: Adjust team goals or values to reflect open mindedness towards AI.
• Emphasize the why: The performance of your AI Agent for customer service is critical for the success and experience of the entire team.

As the Agent absorbs more and more of your support volume, the metrics you use to measure performance will start to change, and you’ll need to set new benchmarks and targets.

Output per person will drop, for instance, which is often the opposite of what most leaders expect. At Fin, our cases handled per hour fell by up to 45% as automation rose.

The remaining work is more demanding, so each case takes longer. Make sure to adjust your goals to address this and communicate it up the ladder. A CFO or CEO who still expects output per person to climb needs to see why the benchmark has changed.

Finally, as your teams take on new areas of ownership, make that visible by recognizing when they’re practicing the right behaviors. In my experience, it’s far more motivating than calling out the gaps. At Fin, the support team runs a monthly “ownership award.” Peers nominate each other, leadership picks a winner, and there's a small prize.

Nobody knows for certain what’s coming next, but that’s not a reason to say nothing. I always advise, even if you don’t know the plan yet, talk to your teams. Be honest about what happens to extra capacity when AI handles the majority of your support volume, where they’ll be focusing their time, and the expectations of them as their roles change.

Change can be destabilizing if people don’t feel like they’re part of it. Involve your teams in the AI rollout, tell them what you do and don’t know, and tell them often.