3 ways Hex makes leading a support team less stressful

If you lead a support team, you’re making calls every week that should be grounded in data — spike diagnosis, ramping new hires, and capacity modeling. But when the analytics aren’t ready, or your data team is slammed, you’re forced into gut decisions.

Hex’s AI tools, like Notebook Agent and Threads, let you self-serve the answers using plain English, and it’s been a game-changer for me as the Product Expert Lead at Hex. You can ask questions in natural language against a trusted model and get defensible outputs that you can share with leadership, all on your own. No Python required, no waiting for a report.

Here are a few ways that I’ve used AI to guide my support team’s decisions that may inspire you too in this era of agentic analytics!

3 ways that Hex’s AI can make your support team more efficient (and your job less stressful)

1. Understand a support ticket spike mystery in minutes

After I got back from vacation, I saw that daily ticket volume had jumped up about 30% from our average while I was gone. Cue the stomach drop — what did I miss?! Why were there so many customer inquiries? I asked Threads, Hex's conversational interface for getting answers from data, a plain-English question:

Why did support ticket volume spike in the past two weeks compared to the prior three months?

Minutes later, I had a crisp insight from the agent’s analysis: The biggest contributor was a large customer that onboarded 300+ new editors to Hex and, as expected, they needed some help from the team to get up to speed! A member of our Customer Engineering team was already two steps ahead and had scheduled live enablement to accelerate their learning.

I quickly quantified “temporary vs. trend,” adjusted our forecasts, and told the team: this isn’t the new normal. I was able to arrive at that conclusion without spreadsheet gymnastics or SQL rabbit holes. Threads let me get immediate visibility, enabling me to de-escalate, keep SLAs intact, and move on.

2. Ramp new hires through thoughtful opportunity identification

Bringing new teammates up to speed isn’t just about shifting tickets around. It’s about identifying the right opportunities for them to learn, contribute, and build confidence.

I recently had two new West Coast hires ready to own their first enterprise accounts, and an overloaded East Coast teammate happy to transition ownership, but we struggled to identify which accounts would be best suited.

I explained the situation to the Threads like I would to a person:

One of my East Coast team members [name redacted] is overloaded. Two new hires on the West Coast were ready for their first dedicated accounts. Which accounts should move? Please consider both support volume by timezone and complexity.

It didn’t just count tickets. It surfaced:

• Accounts spiking outside the current rep’s timezone

• Customers with complex deployments of Hex (a bad fit for new hires)

• Considerations around customer tenure and renewals that may indicate it’s not the best time for them to change ownership

These were details that I might have missed diving through spreadsheets. What would have taken me hours of tedious ticket analysis and VLOOKUPs turned into a few clicks. The result was a calmer queue, cleaner handoffs, and a happier team.

I’ve also been working with Threads as a partner going into 1:1 time with my new team members to understand how they are ramping:

I lead [name redacted] and want to bring meaningful conversation to our 1:1. From looking at their support tickets this week vs. their historical tickets, which product areas have they encountered for the first time recently? Which product areas have they not been exposed to yet? Anything else I should know about their recent support interactions?

This quick prompt helps me not only have relevant, detailed coaching conversations with each individual but also gives me the insight needed to find them new opportunities to continue deepening their product knowledge.

3. Get your headcount request approved

The biggest win I had with our own AI, though? Earning a headcount request with an analysis that I did using the Notebook Agent.

Now, before I continue, you might be asking: why did I use the Notebook Agent here instead of Threads? The main reason was that I wanted to build a capacity model that leadership and I could explore together and continue to use for years to come. The Notebook Agent is better for building lasting analyses to share with others, while Threads is best for one-off investigations and quick questions.

But I digress...

I built a notebook analysis and recorded the insights in a Loom for finance to walk through my case of why we needed extra headcount. It was nothing fancy — no pristine slides — just the work. With my analysis, I was able to show them:

• Current load of the team: Who was over capacity, who was ramping, and which accounts were too complex/noisy for new hires when it came to rebalancing.

• After-hours demand: A sizable share of tickets opened outside West Coast support hours that has been growing over time, creating real coverage gaps for a bi-coastal team.

• Recent spike context: A temporary vs. trend report so we didn’t staff to a blip and could better understand how support demand grows alongside our customer base.

The notebook analysis is what got leadership leaning in — they spun up their own model using the agent in the same notebook. Great! Now we were aligned on definitions and debating assumptions vs. outcomes in the same language.

From there, their follow-up questions got a lot more concrete and I was able to answer them with the help of the Notebook Agent:

• Leadership: “Can rebalancing the team’s workload cover this?”

  I re-ran scenarios with the agent in seconds: moving specific accounts, modeling time-zone coverage, and keeping complex single-tenant accounts with experienced reps. The answer was no.

• Leadership: “Are we staffing to a spike?”

  I replayed the analysis isolating onboarding + transient bugs and showed the baseline trend after those effects. The answer was no. I also explained how onboarding spikes will continue to be expected as new customers get their hands on Hex.

• Leadership: "Show the tradeoffs with new headcount.”

  I compared status quo vs. rebalancing vs. hire, highlighting expected queue stability and after-hours coverage for each.

Because I could rerun any step in seconds, we iterated together. The outcome was a repeatable, inspectable path to justify the next one using the same semantics and definitions.

The context behind the agents

Let me start by saying, you don't need semantic models to get value from Hex's agents, but it certainly helps! In my case, I partnered with our data team on modeling a few metrics I know I'd use frequently.

Hex’s Semantic Model Authoring makes this easy. Instead of hoping everyone interprets a metric the same way, you encode your business logic once — measures, dimensions, joins, and rules — and Hex’s AI (Notebook Agent + Threads) answers questions in those semantics. Here are some examples of what I partnered with the data team on to model, as the owner of those business definitions.

Context that I partnered with the data team on to ensure accuracy:

• Measures: tickets_opened, after_hours_rate, backlog_age, time_to_first_response.

• Dimensions: account_tier, onboarding_phase, time_zone, product_area, rep_tenure

• Rules/logic: what counts as an “outage,” how to classify single-tenant complexity, which tickets belong to dedicated vs. round-robin queues.

With those in place, I can ask plain-English questions — “Are we staffing to a temporary spike or a trend?” “Which accounts should move to balance time zones?” — and get answers grounded in the same definitions leaders see.

You don't need to be technical to transform support through data

Using Semantic Authoring with AI, I can have data-driven conversations about our support operations without writing complex code. Threads and Notebook Agent translate my questions into answers with a data trail I can click into, helping me make informed decisions quickly.

If you don't want to wait for data requests, partner with your data team to create a semantic model and then jump into Hex to ask questions yourself! It's fun and the impact is immediate with data-backed recommendations that leadership can trust.